United States v. Booker

543 U.S. 220 (2005)

UNITED STATES
v.
BOOKER

No. 04-104.

Supreme Court of United States.

Argued October 4, 2004. Decided January 12, 2005.^[*] CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE SEVENTH CIRCUIT.

STEVENS, J., delivered the opinion of the Court in part, in which SCALIA, SOUTER, THOMAS, and GINSBURG, JJ., joined. BREYER, J., delivered the opinion of the Court in part, in which REHNQUIST, C. J., and O'CONNOR, KENNEDY, and GINSBURG, JJ., joined, post, p. 244. STEVENS, J., filed an opinion dissenting in part, in which SOUTER, J., joined, and in which SCALIA, J., joined except for Part III and footnote 17, post, p. 272. SCALIA, J., post, p. 303, and THOMAS, J., post, p. 313, filed opinions dissenting in part. BREYER, J., filed an opinion dissenting in part, in which REHNQUIST, C. J., and O'CONNOR and KENNEDY, JJ., joined, post, p. 326.

Acting Solicitor General Clement argued the cause for the United States in both cases. With him on the brief were Assistant Attorney General Wray, Deputy Solicitor General Dreeben, James A. Feldman, Dan Himmelfarb, and Nina Goodman.

T. Christopher Kelly argued the cause for respondent in No. 04-104. With him on the brief was Dean A. Strang. Rosemary Curran Scapicchio argued the cause for respondent in No. 04-105. With her on the brief were Carter G. Phillips, Jeffrey T. Green, Eric A. Shumsky, and Martin G. Weinberg.^[†]

JUSTICE STEVENS delivered the opinion of the Court in part.^[*]

The question presented in each of these cases is whether an application of the Federal Sentencing Guidelines violated the Sixth Amendment. In each case, the courts below held that binding rules set forth in the Guidelines limited the severity of the sentence that the judge could lawfully impose on the defendant based on the facts found by the jury at his trial. In both cases the courts rejected, on the basis of our decision in Blakely v. Washington, 542 U. S. 296 (2004), the Government's recommended application of the Sentencing Guidelines because the proposed sentences were based on additional facts that the sentencing judge found by a preponderance of the evidence. We hold that both courts correctly concluded that the Sixth Amendment as construed in Blakely does apply to the Sentencing Guidelines. In a separate opinion authored by JUSTICE BREYER, the Court concludes that in light of this holding, two provisions of the Sentencing Reform Act of 1984 (SRA) that have the effect of making the Guidelines mandatory must be invalidated in order to allow the statute to operate in a manner consistent with congressional intent.

I

Respondent Booker was charged with possession with intent to distribute at least 50 grams of cocaine base (crack). Having heard evidence that he had 92.5 grams in his duffel bag, the jury found him guilty of violating 21 U. S. C. § 841(a)(1). That statute prescribes a minimum sentence of 10 years in prison and a maximum sentence of life for that offense. § 841(b)(1)(A)(iii).

Based upon Booker's criminal history and the quantity of drugs found by the jury, the Sentencing Guidelines required the District Court Judge to select a "base" sentence of not less than 210 nor more than 262 months in prison. See United States Sentencing Commission, Guidelines Manual §§ 2D1.1(c)(4), 4A1.1 (Nov. 2003) (USSG). The judge, however, held a post-trial sentencing proceeding and concluded by a preponderance of the evidence that Booker had possessed an additional 566 grams of crack and that he was guilty of obstructing justice. Those findings mandated that the judge select a sentence between 360 months and life imprisonment; the judge imposed a sentence at the low end of the range. Thus, instead of the sentence of 21 years and 10 months that the judge could have imposed on the basis of the facts proved to the jury beyond a reasonable doubt, Booker received a 30-year sentence.

Over the dissent of Judge Easterbrook, the Court of Appeals for the Seventh Circuit held that this application of the Sentencing Guidelines conflicted with our holding in Apprendi v. New Jersey, 530 U. S. 466, 490 (2000), that "[o]ther than the fact of a prior conviction, any fact that increases the penalty for a crime beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt." 375 F. 3d 508, 510 (2004). The majority relied on our holding in Blakely, 542 U. S. 296, that "the `statutory maximum' for Apprendi purposes is the maximum sentence a judge may impose solely on the basis of the facts reflected in the jury verdict or admitted by the defendant." Id., at 303. The court held that the sentence violated the Sixth Amendment, and remanded with instructions to the District Court either to sentence respondent within the sentencing range supported by the jury's findings or to hold a separate sentencing hearing before a jury.

Respondent Fanfan was charged with conspiracy to distribute and to possess with intent to distribute at least 500 grams of cocaine in violation of 21 U. S. C. §§ 846, 841(a)(1), and 841(b)(1)(B)(ii). He was convicted by the jury after it answered "Yes" to the question "Was the amount of cocaine 500 or more grams?" App. C to Pet. for Cert. in No. 04-105, p. 15a. Under the Guidelines, without additional findings of fact, the maximum sentence authorized by the jury verdict was imprisonment for 78 months.

A few days after our decision in Blakely, the trial judge conducted a sentencing hearing at which he found additional facts that, under the Guidelines, would have authorized a sentence in the 188-to-235-month range. Specifically, he found that respondent Fanfan was responsible for 2.5 kilograms of cocaine powder, and 261.6 grams of crack. He also concluded that respondent had been an organizer, leader, manager, or supervisor in the criminal activity. Both findings were made by a preponderance of the evidence. Under the Guidelines, these additional findings would have required an enhanced sentence of 15 or 16 years instead of the 5 or 6 years authorized by the jury verdict alone. Relying not only on the majority opinion in Blakely, but also on the categorical statements in the dissenting opinions and in the Solicitor General's brief in Blakely, see App. A to Pet. for Cert. in No. 04-105, pp. 6a-7a, the judge concluded that he could not follow the particular provisions of the Sentencing Guidelines "which involve drug quantity and role enhancement," id., at 11a. Expressly refusing to make "any blanket decision about the federal guidelines," he followed the provisions of the Guidelines that did not implicate the Sixth Amendment by imposing a sentence on respondent "based solely upon the jury verdict in this case." Ibid.

Following the denial of its motion to correct the sentence in Fanfan's case, the Government filed a notice of appeal in the Court of Appeals for the First Circuit, and a petition in this Court for a writ of certiorari before judgment. Because of the importance of the questions presented, we granted that petition, 542 U. S. 956 (2004), as well as a similar petition filed by the Government in Booker's case, ibid. In both petitions, the Government asks us to determine whether our Apprendi line of cases applies to the Sentencing Guidelines, and if so, what portions of the Guidelines remain in effect.^[1]

In this opinion, we explain why we agree with the lower courts' answer to the first question. In a separate opinion for the Court, JUSTICE BREYER explains the Court's answer to the second question.

II

It has been settled throughout our history that the Constitution protects every criminal defendant "against conviction except upon proof beyond a reasonable doubt of every fact necessary to constitute the crime with which he is charged." In re Winship, 397 U. S. 358, 364 (1970). It is equally clear that the "Constitution gives a criminal defendant the right to demand that a jury find him guilty of all the elements of the crime with which he is charged." United States v. Gaudin, 515 U. S. 506, 511 (1995). These basic precepts, firmly rooted in the common law, have provided the basis for recent decisions interpreting modern criminal statutes and sentencing procedures.

In Jones v. United States, 526 U. S. 227, 230 (1999), we considered the federal carjacking statute, which provides three different maximum sentences depending on the extent of harm to the victim: 15 years in jail if there was no serious injury to a victim, 25 years if there was "serious bodily injury," and life in prison if death resulted. 18 U. S. C. § 2119 (1988 ed., Supp. V). In spite of the fact that the statute "at first glance has a look to it suggesting [that the provisions relating to the extent of harm to the victim] are only sentencing provisions," 526 U. S., at 232, we concluded that the harm to the victim was an element of the crime. That conclusion was supported by the statutory text and structure, and was influenced by our desire to avoid the constitutional issues implicated by a contrary holding, which would have reduced the jury's role "to the relative importance of low-level gate-keeping." Id., at 244. Foreshadowing the result we reach today, we noted that our holding was consistent with a "rule requiring jury determination of facts that raise a sentencing ceiling" in state and federal sentencing guidelines systems. Id., at 251-252, n. 11.

In Apprendi v. New Jersey, 530 U. S. 466 (2000), the defendant pleaded guilty to second-degree possession of a firearm for an unlawful purpose, which carried a prison term of 5-to-10 years. Thereafter, the trial court found that his conduct had violated New Jersey's "hate crime" law because it was racially motivated, and imposed a 12-year sentence. This Court set aside the enhanced sentence. We held: "Other than the fact of a prior conviction, any fact that increases the penalty for a crime beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt." Id., at 490.

The fact that New Jersey labeled the hate crime a "sentence enhancement" rather than a separate criminal act was irrelevant for constitutional purposes. Id., at 478. As a matter of simple justice, it seemed obvious that the procedural safeguards designed to protect Apprendi from punishment for the possession of a firearm should apply equally to his violation of the hate crime statute. Merely using the label "sentence enhancement" to describe the latter did not provide a principled basis for treating the two crimes differently. Id., at 476.

In Ring v. Arizona, 536 U. S. 584 (2002), we reaffirmed our conclusion that the characterization of critical facts is constitutionally irrelevant. There, we held that it was impermissible for "the trial judge, sitting alone" to determine the presence or absence of the aggravating factors required by Arizona law for imposition of the death penalty. Id., at 588-589. "If a State makes an increase in a defendant's authorized punishment contingent on the finding of a fact, that fact—no matter how the State labels it—must be found by a jury beyond a reasonable doubt." Id., at 602. Our opinion made it clear that ultimately, while the procedural error in Ring's case might have been harmless because the necessary finding was implicit in the jury's guilty verdict, id., at 609, n. 7, "the characterization of a fact or circumstance as an `element' or a `sentencing factor' is not determinative of the question `who decides,' judge or jury," id., at 605.

In Blakely v. Washington, 542 U. S. 296 (2004), we dealt with a determinate sentencing scheme similar to the Federal Sentencing Guidelines. There the defendant pleaded guilty to kidnaping, a class B felony punishable by a term of not more than 10 years. Other provisions of Washington law, comparable to the Federal Sentencing Guidelines, mandated a "standard" sentence of 49-to-53 months, unless the judge found aggravating facts justifying an exceptional sentence. Although the prosecutor recommended a sentence in the standard range, the judge found that the defendant had acted with "`deliberate cruelty'" and sentenced him to 90 months. Id., at 300.

For reasons explained in Jones, Apprendi, and Ring, the requirements of the Sixth Amendment were clear. The application of Washington's sentencing scheme violated the defendant's right to have the jury find the existence of "`any particular fact'" that the law makes essential to his punishment. 542 U. S., at 301. That right is implicated whenever a judge seeks to impose a sentence that is not solely based on "facts reflected in the jury verdict or admitted by the defendant." Id., at 303 (emphasis deleted). We rejected the State's argument that the jury verdict was sufficient to authorize a sentence within the general 10-year sentence for class B felonies, noting that under Washington law, the judge was required to find additional facts in order to impose the greater 90-month sentence. Our precedents, we explained, make clear "that the `statutory maximum' for Apprendi purposes is the maximum sentence a judge may impose solely on the basis of the facts reflected in the jury verdict or admitted by the defendant." Ibid. (emphasis in original). The determination that the defendant acted with deliberate cruelty, like the determination in Apprendi that the defendant acted with racial malice, increased the sentence that the defendant could have otherwise received. Since this fact was found by a judge using a preponderance of the evidence standard, the sentence violated Blakely's Sixth Amendment rights.

As the dissenting opinions in Blakely recognized, there is no distinction of constitutional significance between the Federal Sentencing Guidelines and the Washington procedures at issue in that case. See, e. g., 542 U. S., at 325 (opinion of O'CONNOR, J.) ("The structure of the Federal Guidelines likewise does not, as the Government half-heartedly suggests, provide any grounds for distinction. . . . If anything, the structural differences that do exist make the Federal Guidelines more vulnerable to attack"). This conclusion rests on the premise, common to both systems, that the relevant sentencing rules are mandatory and impose binding requirements on all sentencing judges.

If the Guidelines as currently written could be read as merely advisory provisions that recommended, rather than required, the selection of particular sentences in response to differing sets of facts, their use would not implicate the Sixth Amendment. We have never doubted the authority of a judge to exercise broad discretion in imposing a sentence within a statutory range. See Apprendi, 530 U. S., at 481; Williams v. New York, 337 U. S. 241, 246 (1949). Indeed, everyone agrees that the constitutional issues presented by these cases would have been avoided entirely if Congress had omitted from the SRA the provisions that make the Guidelines binding on district judges; it is that circumstance that makes the Court's answer to the second question presented possible. For when a trial judge exercises his discretion to select a specific sentence within a defined range, the defendant has no right to a jury determination of the facts that the judge deems relevant.

The Guidelines as written, however, are not advisory; they are mandatory and binding on all judges.^[2] While subsection (a) of § 3553 of the sentencing statute^[3] lists the Sentencing Guidelines as one factor to be considered in imposing a sentence, subsection (b) directs that the court "shall impose a sentence of the kind, and within the range" established by the Guidelines, subject to departures in specific, limited cases. (Emphasis added.) Because they are binding on judges, we have consistently held that the Guidelines have the force and effect of laws. See, e. g., Mistretta v. United States, 488 U. S. 361, 391 (1989); Stinson v. United States, 508 U. S. 36, 42 (1993).

The availability of a departure in specified circumstances does not avoid the constitutional issue, just as it did not in Blakely itself. The Guidelines permit departures from the prescribed sentencing range in cases in which the judge "finds that there exists an aggravating or mitigating circumstance of a kind, or to a degree, not adequately taken into consideration by the Sentencing Commission in formulating the guidelines that should result in a sentence different from that described." 18 U. S. C. A. § 3553(b)(1) (Supp. 2004). At first glance, one might believe that the ability of a district judge to depart from the Guidelines means that she is bound only by the statutory maximum. Were this the case, there would be no Apprendi problem. Importantly, however, departures are not available in every case, and in fact are unavailable in most. In most cases, as a matter of law, the Commission will have adequately taken all relevant factors into account, and no departure will be legally permissible. In those instances, the judge is bound to impose a sentence within the Guidelines range. It was for this reason that we rejected a similar argument in Blakely, holding that although the Washington statute allowed the judge to impose a sentence outside the sentencing range for "`substantial and compelling reasons,'" that exception was not available for Blakely himself. 542 U. S., at 299. The sentencing judge would have been reversed had he invoked the departure section to justify the sentence.

Booker's case illustrates the mandatory nature of the Guidelines. The jury convicted him of possessing at least 50 grams of crack in violation of 21 U. S. C. § 841(b)(1)(A)(iii) based on evidence that he had 92.5 grams of crack in his duffel bag. Under these facts, the Guidelines specified an offense level of 32, which, given the defendant's criminal history category, authorized a sentence of 210-to-262 months. See USSG § 2D1.1(c)(4). Booker's is a run-of-the-mill drug case, and does not present any factors that were inadequately considered by the Commission. The sentencing judge would therefore have been reversed had he not imposed a sentence within the level 32 Guidelines range.

Booker's actual sentence, however, was 360 months, almost 10 years longer than the Guidelines range supported by the jury verdict alone. To reach this sentence, the judge found facts beyond those found by the jury: namely, that Booker possessed 566 grams of crack in addition to the 92.5 grams in his duffel bag. The jury never heard any evidence of the additional drug quantity, and the judge found it true by a preponderance of the evidence. Thus, just as in Blakely, "the jury's verdict alone does not authorize the sentence. The judge acquires that authority only upon finding some additional fact." 542 U. S., at 305. There is no relevant distinction between the sentence imposed pursuant to the Washington statutes in Blakely and the sentences imposed pursuant to the Federal Sentencing Guidelines in these cases.

In his dissent, post, at 327-329, JUSTICE BREYER argues on historical grounds that the Guidelines scheme is constitutional across the board. He points to traditional judicial authority to increase sentences to take account of any unusual blameworthiness in the manner employed in committing a crime, an authority that the Guidelines require to be exercised consistently throughout the system. This tradition, however, does not provide a sound guide to enforcement of the Sixth Amendment's guarantee of a jury trial in today's world.

It is quite true that once determinate sentencing had fallen from favor, American judges commonly determined facts justifying a choice of a heavier sentence on account of the manner in which particular defendants acted. Apprendi, 530 U. S., at 481. In 1986, however, our own cases first recognized a new trend in the legislative regulation of sentencing when we considered the significance of facts selected by legislatures that not only authorized, or even mandated, heavier sentences than would otherwise have been imposed, but increased the range of sentences possible for the underlying crime. See McMillan v. Pennsylvania, 477 U. S. 79, 87-88 (1986). Provisions for such enhancements of the permissible sentencing range reflected growing and wholly justified legislative concern about the proliferation and variety of drug crimes and their frequent identification with firearms offenses.

The effect of the increasing emphasis on facts that enhanced sentencing ranges, however, was to increase the judge's power and diminish that of the jury. It became the judge, not the jury, who determined the upper limits of sentencing, and the facts determined were not required to be raised before trial or proved by more than a preponderance.

As the enhancements became greater, the jury's finding of the underlying crime became less significant. And the enhancements became very serious indeed. See, e. g., Jones, 526 U. S., at 230-231 (judge's finding increased the maximum sentence from 15 to 25 years); respondent Booker's case (from 262 months to a life sentence); respondent Fanfan's case (from 78 to 235 months); United States v. Rodriguez, 73 F. 3d 161, 162-163 (CA7 1996) (Posner, C. J., dissenting from denial of rehearing en banc) (from approximately 54 months to a life sentence); United States v. Hammoud, 381 F. 3d 316, 361-362 (CA4 2004) (en banc) (Motz, J., dissenting) (actual sentence increased from 57 months to 155 years).

As it thus became clear that sentencing was no longer taking place in the tradition that JUSTICE BREYER invokes, the Court was faced with the issue of preserving an ancient guarantee under a new set of circumstances. The new sentencing practice forced the Court to address the question how the right of jury trial could be preserved, in a meaningful way guaranteeing that the jury would still stand between the individual and the power of the government under the new sentencing regime. And it is the new circumstances, not a tradition or practice that the new circumstances have superseded, that have led us to the answer first considered in Jones and developed in Apprendi and subsequent cases culminating with this one. It is an answer not motivated by Sixth Amendment formalism, but by the need to preserve Sixth Amendment substance.

III

The Government advances three arguments in support of its submission that we should not apply our reasoning in Blakely to the Federal Sentencing Guidelines. It contends that Blakely is distinguishable because the Guidelines were promulgated by a Commission rather than the Legislature; that principles of stare decisis require us to follow four earlier decisions that are arguably inconsistent with Blakely; and that the application of Blakely to the Guidelines would conflict with separation-of-powers principles reflected in Mistretta v. United States, 488 U. S. 361 (1989). These arguments are unpersuasive.

Commission v. Legislature:

In our judgment the fact that the Guidelines were promulgated by the Sentencing Commission, rather than Congress, lacks constitutional significance. In order to impose the defendants' sentences under the Guidelines, the judges in these cases were required to find an additional fact, such as drug quantity, just as the judge found the additional fact of serious bodily injury to the victim in Jones. As far as the defendants are concerned, they face significantly higher sentences — in Booker's case almost 10 years higher — because a judge found true by a preponderance of the evidence a fact that was never submitted to the jury. Regardless of whether Congress or a Sentencing Commission concluded that a particular fact must be proved in order to sentence a defendant within a particular range, "[t]he Framers would not have thought it too much to demand that, before depriving a man of [ten] more years of his liberty, the State should suffer the modest inconvenience of submitting its accusation to `the unanimous suffrage of twelve of his equals and neighbours,' rather than a lone employee of the State." Blakely, 542 U. S., at 313-314 (citation omitted).

The Government correctly notes that in Apprendi we referred to "`any fact that increases the penalty for a crime beyond the prescribed statutory maximum . . . .'" Brief for United States 15 (quoting Apprendi, 530 U. S., at 490 (emphasis in Brief for United States)). The simple answer, of course, is that we were only considering a statute in that case; we expressly declined to consider the Guidelines. See Apprendi, 530 U. S., at 497, n. 21. It was therefore appropriate to state the rule in that case in terms of a "statutory maximum" rather than answering a question not properly before us.

More important than the language used in our holding in Apprendi are the principles we sought to vindicate. Those principles are unquestionably applicable to the Guidelines. They are not the product of recent innovations in our jurisprudence, but rather have their genesis in the ideals our constitutional tradition assimilated from the common law. See Jones, 526 U. S., at 244-248. The Framers of the Constitution understood the threat of "judicial despotism" that could arise from "arbitrary punishments upon arbitrary convictions" without the benefit of a jury in criminal cases. The Federalist No. 83, p. 499 (C. Rossiter ed. 1961) (A. Hamilton). The Founders presumably carried this concern from England, in which the right to a jury trial had been enshrined since the Magna Carta. As we noted in Apprendi:

"[T]he historical foundation for our recognition of these principles extends down centuries into the common law. `[T]o guard against a spirit of oppression and tyranny on the part of rulers,' and `as the great bulwark of [our] civil and political liberties,' trial by jury has been understood to require that `the truth of every accusation, whether preferred in the shape of indictment, information, or appeal, should afterwards be confirmed by the unanimous suffrage of twelve of [the defendant's] equals and neighbours . . . .'" 530 U. S., at 477 (citations omitted).

Regardless of whether the legal basis of the accusation is in a statute or in guidelines promulgated by an independent commission, the principles behind the jury trial right are equally applicable.

Stare Decisis:

The Government next argues that four recent cases preclude our application of Blakely to the Sentencing Guidelines. We disagree. In United States v. Dunnigan, 507 U. S. 87 (1993), we held that the provisions of the Guidelines that require a sentence enhancement if the judge determines that the defendant committed perjury do not violate the privilege of the accused to testify on her own behalf. There was no contention that the enhancement was invalid because it resulted in a more severe sentence than the jury verdict had authorized. Accordingly, we found this case indistinguishable from United States v. Grayson, 438 U. S. 41 (1978), a pre-Guidelines case in which we upheld a similar sentence increase. Applying Blakely to the Guidelines would invalidate a sentence that relied on such an enhancement if the resulting sentence was outside the range authorized by the jury verdict. Nevertheless, there are many situations in which the district judge might find that the enhancement is warranted, yet still sentence the defendant within the range authorized by the jury. See post, at 276-279 (STEVENS, J., dissenting in part). Thus, while the reach of Dunnigan may be limited, we need not overrule it.

In Witte v. United States, 515 U. S. 389 (1995), we held that the Double Jeopardy Clause did not bar a prosecution for conduct that had provided the basis for an enhancement of the defendant's sentence in a prior case. "We concluded that `consideration of information about the defendant's character and conduct at sentencing does not result in "punishment" for any offense other than the one of which the defendant was convicted.' Rather, the defendant is `punished only for the fact that the present offense was carried out in a manner that warrants increased punishment . . . .'" United States v. Watts, 519 U. S. 148, 155 (1997) (per curiam) (emphasis deleted) (quoting Witte, 515 U. S., at 401, 403). In Watts, relying on Witte, we held that the Double Jeopardy Clause permitted a court to consider acquitted conduct in sentencing a defendant under the Guidelines. In neither Witte nor Watts was there any contention that the sentencing enhancement had exceeded the sentence authorized by the jury verdict in violation of the Sixth Amendment. The issue we confront today simply was not presented.^[4]

Finally, in Edwards v. United States, 523 U. S. 511 (1998), the Court held that a jury's general verdict finding the defendants guilty of a conspiracy involving either cocaine or crack supported a sentence based on their involvement with both drugs. Even though the indictment had charged that their conspiracy embraced both, they argued on appeal that the verdict limited the judge's sentencing authority. We recognized that the defendants' statutory and constitutional claims might have had merit if it had been possible to argue that their crack-related activities were not part of the same conspiracy as their cocaine activities. But they failed to make that argument, and, based on our review of the record which showed "a series of interrelated drug transactions involving both cocaine and crack," we concluded that no such claim could succeed.^[5]Id., at 515.

None of our prior cases is inconsistent with today's decision. Stare decisis does not compel us to limit Blakely's holding.

Separation of Powers:

Finally, the Government and, to a lesser extent, JUSTICE BREYER's dissent, argue that any holding that would require Guidelines sentencing factors to be proved to a jury beyond a reasonable doubt would effectively transform them into a code defining elements of criminal offenses. The result, according to the Government, would be an unconstitutional grant to the Sentencing Commission of the inherently legislative power to define criminal elements.

There is no merit to this argument because the Commission's authority to identify the facts relevant to sentencing decisions and to determine the impact of such facts on federal sentences is precisely the same whether one labels such facts "sentencing factors" or "elements" of crimes. Our decision in Mistretta, 488 U. S., at 371, upholding the validity of the delegation of that authority, is unaffected by the characterization of such facts, or by the procedures used to find such facts in particular sentencing proceedings. Indeed, we rejected a similar argument in Jones:

"Contrary to the dissent's suggestion, the constitutional proposition that drives our concern in no way `call[s] into question the principle that the definition of the elements of a criminal offense is entrusted to the legislature.' The constitutional guarantees that give rise to our concern in no way restrict the ability of legislatures to identify the conduct they wish to characterize as criminal or to define the facts whose proof is essential to the establishment of criminal liability. The constitutional safeguards that figure in our analysis concern not the identity of the elements defining criminal liability but only the required procedures for finding the facts that determine the maximum permissible punishment; these are the safeguards going to the formality of notice, the identity of the factfinder, and the burden of proof." 526 U. S., at 243, n. 6 (citation omitted).

Our holding today does not call into question any aspect of our decision in Mistretta. That decision was premised on an understanding that the Commission, rather than performing adjudicatory functions, instead makes political and substantive decisions. 488 U. S., at 393. We noted that the promulgation of the Guidelines was much like other activities in the Judicial Branch, such as the creation of the Federal Rules of Evidence, all of which are nonadjudicatory activities. Id., at 387. We also noted that "Congress may delegate to the Judicial Branch nonadjudicatory functions that do not trench upon the prerogatives of another Branch and that are appropriate to the central mission of the Judiciary." Id., at 388. While we recognized that the Guidelines were more substantive than the Rules of Evidence or other nonadjudicatory functions delegated to the Judicial Branch, we nonetheless concluded that such a delegation did not exceed Congress' powers.

Further, a recognition that the Commission did not exercise judicial authority, but was more properly thought of as exercising some sort of legislative power, ibid., was essential to our holding. If the Commission in fact performed adjudicatory functions, it would have violated Article III because some of the members were not Article III judges. As we explained:

"[T]he `practical consequences' of locating the Commission within the Judicial Branch pose no threat of undermining the integrity of the Judicial Branch or of expanding the powers of the Judiciary beyond constitutional bounds by uniting within the Branch the political or quasi-legislative power of the Commission with the judicial power of the courts. [The Commission's] powers are not united with the powers of the Judiciary in a way that has meaning for separation-of-powers analysis. Whatever constitutional problems might arise if the powers of the Commission were vested in a court, the Commission is not a court, does not exercise judicial power, and is not controlled by or accountable to members of the Judicial Branch." Id., at 393.

We have thus always recognized the fact that the Commission is an independent agency that exercises policymaking authority delegated to it by Congress. Nothing in our holding today is inconsistent with our decision in Mistretta.

IV

All of the foregoing supports our conclusion that our holding in Blakely applies to the Sentencing Guidelines. We recognize, as we did in Jones, Apprendi, and Blakely, that in some cases jury factfinding may impair the most expedient and efficient sentencing of defendants. But the interest in fairness and reliability protected by the right to a jury trial — a common-law right that defendants enjoyed for centuries and that is now enshrined in the Sixth Amendment — has always outweighed the interest in concluding trials swiftly. Blakely, 542 U. S., at 313. As Blackstone put it:

"[H]owever convenient these [new methods of trial] may appear at first, (as doubtless all arbitrary powers, well executed, are the most convenient) yet let it be again remembered, that delays, and little inconveniences in the forms of justice, are the price that all free nations must pay for their liberty in more substantial matters; that these inroads upon this sacred bulwark of the nation are fundamentally opposite to the spirit of our constitution; and that, though begun in trifles, the precedent may gradually increase and spread, to the utter disuse of juries in questions of the most momentous concerns." 4 Commentaries on the Laws of England 343-344 (1769).

Accordingly, we reaffirm our holding in Apprendi: Any fact (other than a prior conviction) which is necessary to support a sentence exceeding the maximum authorized by the facts established by a plea of guilty or a jury verdict must be admitted by the defendant or proved to a jury beyond a reasonable doubt.

JUSTICE BREYER delivered the opinion of the Court in part.^[*]

The first question that the Government has presented in these cases is the following:

"Whether the Sixth Amendment is violated by the imposition of an enhanced sentence under the United States Sentencing Guidelines based on the sentencing judge's determination of a fact (other than a prior conviction) that was not found by the jury or admitted by the defendant." Pet. for Cert. in No. 04-104, p. (I).

The Court, in an opinion by JUSTICE STEVENS, answers this question in the affirmative. Applying its decisions in Apprendi v. New Jersey, 530 U. S. 466 (2000), and Blakely v. Washington, 542 U. S. 296 (2004), to the Federal Sentencing Guidelines, the Court holds that, in the circumstances mentioned, the Sixth Amendment requires juries, not judges, to find facts relevant to sentencing. See ante, at 226-227, 244 (STEVENS, J., opinion of the Court).

We here turn to the second question presented, a question that concerns the remedy. We must decide whether or to what extent, "as a matter of severability analysis," the Guidelines "as a whole" are "inapplicable . . . such that the sentencing court must exercise its discretion to sentence the defendant within the maximum and minimum set by statute for the offense of conviction." Pet. for Cert. in No. 04-104, p. (I).

We answer the question of remedy by finding the provision of the federal sentencing statute that makes the Guidelines mandatory, 18 U. S. C. A. § 3553(b)(1) (Supp. 2004), incompatible with today's constitutional holding. We conclude that this provision must be severed and excised, as must one other statutory section, § 3742(e) (main ed. and Supp. 2004), which depends upon the Guidelines' mandatory nature. So modified, the federal sentencing statute, see Sentencing Reform Act of 1984 (Sentencing Act), as amended, 18 U. S. C. § 3551 et seq., 28 U. S. C. § 991 et seq., makes the Guidelines effectively advisory. It requires a sentencing court to consider Guidelines ranges, see 18 U. S. C. A. § 3553(a)(4) (Supp. 2004), but it permits the court to tailor the sentence in light of other statutory concerns as well, see § 3553(a) (Supp. 2004).

I

We answer the remedial question by looking to legislative intent. See, e. g., Minnesota v. Mille Lacs Band of Chippewa Indians, 526 U. S. 172, 191 (1999); Alaska Airlines, Inc. v. Brock, 480 U. S. 678, 684 (1987); Regan v. Time, Inc., 468 U. S. 641, 653 (1984) (plurality opinion). We seek to determine what "Congress would have intended" in light of the Court's constitutional holding. Denver Area Ed. Telecommunications Consortium, Inc. v. FCC, 518 U. S. 727, 767 (1996) (plurality opinion) ("Would Congress still have passed" the valid sections "had it known" about the constitutional invalidity of the other portions of the statute? (internal quotation marks omitted)). In this instance, we must determine which of the two following remedial approaches is the more compatible with the Legislature's intent as embodied in the 1984 Sentencing Act.

One approach, that of JUSTICE STEVENS' dissent, would retain the Sentencing Act (and the Guidelines) as written, but would engraft onto the existing system today's Sixth Amendment "jury trial" requirement. The addition would change the Guidelines by preventing the sentencing court from increasing a sentence on the basis of a fact that the jury did not find (or that the offender did not admit).

The other approach, which we now adopt, would (through severance and excision of two provisions) make the Guidelines system advisory while maintaining a strong connection between the sentence imposed and the offender's real conduct — a connection important to the increased uniformity of sentencing that Congress intended its Guidelines system to achieve.

Both approaches would significantly alter the system that Congress designed. But today's constitutional holding means that it is no longer possible to maintain the judicial factfinding that Congress thought would underpin the mandatory Guidelines system that it sought to create and that Congress wrote into the Act in 18 U. S. C. A. §§ 3553(a) and 3661 (main ed. and Supp. 2004). Hence we must decide whether we would deviate less radically from Congress' intended system (1) by superimposing the constitutional requirement announced today or (2) through elimination of some provisions of the statute.

To say this is not to create a new kind of severability analysis. Post, at 291 (STEVENS, J., dissenting in part). Rather, it is to recognize that sometimes severability questions (questions as to how, or whether, Congress would intend a statute to apply) can arise when a legislatively unforeseen constitutional problem requires modification of a statutory provision as applied in a significant number of instances. Compare, e. g., Welsh v. United States, 398 U. S. 333, 361 (1970) (Harlan, J., concurring in result) (explaining that when a statute is defective because of its failure to extend to some group a constitutionally required benefit, the court may "either declare it a nullity" or "extend" the benefit "to include those who are aggrieved by exclusion"); Heckler v. Mathews, 465 U. S. 728, 739, n. 5 (1984) ("Although . . . ordinarily `extension, rather than nullification, is the proper course,' the court should not, of course, `use its remedial powers to circumvent the intent of the legislature . . . .'" (quoting Califano v. Westcott, 443 U. S. 76, 89 (1979) and id., at 94 (Powell, J., concurring in part and dissenting in part))); Sloan v. Lemon, 413 U. S. 825, 834 (1973) (striking down entire Pennsylvania tuition reimbursement statute because to eliminate only unconstitutional applications "would be to create a program quite different from the one the legislature actually adopted"). See also post, at 320, 323 (THOMAS, J., dissenting in part) ("[S]everability questions" can "arise from unconstitutional applications" of statutes, and such a question "is squarely presented" here); Vermeule, Saving Constructions, 85 Geo. L. J. 1945, 1950, n. 26 (1997).

In today's context — a highly complex statute, interrelated provisions, and a constitutional requirement that creates fundamental change — we cannot assume that Congress, if faced with the statute's invalidity in key applications, would have preferred to apply the statute in as many other instances as possible. Neither can we determine likely congressional intent mechanically. We cannot simply approach the problem grammatically, say, by looking to see whether the constitutional requirement and the words of the Act are linguistically compatible.

Nor do simple numbers provide an answer. It is, of course, true that the numbers show that the constitutional jury trial requirement would lead to additional decisionmaking by juries in only a minority of cases. See post, at 277 (STEVENS, J., dissenting in part). Prosecutors and defense attorneys would still resolve the lion's share of criminal matters through plea bargaining, and plea bargaining takes place without a jury. See ibid. Many of the rest involve only simple issues calling for no upward Guidelines adjustment. See post, at 275. And in at least some of the remainder, a judge may find adequate room to adjust a sentence within the single Guidelines range to which the jury verdict points, or within the overlap between that range and the next highest. See post, at 278-279.

But the constitutional jury trial requirement would nonetheless affect every case. It would affect decisions about whether to go to trial. It would affect the content of plea negotiations. It would alter the judge's role in sentencing. Thus we must determine likely intent not by counting proceedings, but by evaluating the consequences of the Court's constitutional requirement in light of the Act's language, its history, and its basic purposes.

While reasonable minds can, and do, differ about the outcome, we conclude that the constitutional jury trial requirement is not compatible with the Act as written and that some severance and excision are necessary. In Part II, infra, we explain the incompatibility. In Part III, infra, we describe the necessary excision. In Part IV, infra, we explain why we have rejected other possibilities. In essence, in what follows, we explain both (1) why Congress would likely have preferred the total invalidation of the Act to an Act with the Court's Sixth Amendment requirement engrafted onto it, and (2) why Congress would likely have preferred the excision of some of the Act, namely the Act's mandatory language, to the invalidation of the entire Act. That is to say, in light of today's holding, we compare maintaining the Act as written with jury factfinding added (the dissenters' proposed remedy) to the total invalidation of the statute, and conclude that Congress would have preferred the latter. We then compare our own remedy to the total invalidation of the statute, and conclude that Congress would have preferred our remedy.

II

Several considerations convince us that, were the Court's constitutional requirement added onto the Sentencing Act as currently written, the requirement would so transform the scheme that Congress created that Congress likely would not have intended the Act as so modified to stand. First, the statute's text states that "[t]he court" when sentencing will consider "the nature and circumstances of the offense and the history and characteristics of the defendant." 18 U. S. C. A. § 3553(a)(1) (main ed. and Supp. 2004). In context, the words "the court" mean "the judge without the jury," not "the judge working together with the jury." A further statutory provision, by removing typical "jury trial" evidentiary limitations, makes this clear. See § 3661 (ruling out any "limitation . . . on the information concerning the [offender's] background, character, and conduct" that the "court . . . may receive"). The Act's history confirms it. See, e. g., S. Rep. No. 98-225, p. 51 (1983) (the Guidelines system "will guide the judge in making" sentencing decisions (emphasis added)); id., at 52 (before sentencing, "the judge" must consider "the nature and circumstances of the offense"); id., at 53 ("the judge" must conduct "a comprehensive examination of the characteristics of the particular offense and the particular offender").

This provision is tied to the provision of the Act that makes the Guidelines mandatory, see § 3553(b)(1) (Supp. 2004). They are part and parcel of a single, unified whole — a whole that Congress intended to apply to all federal sentencing.

This provision makes it difficult to justify JUSTICE STEVENS' approach, for that approach requires reading the words "the court" as if they meant "the judge working together with the jury." Unlike JUSTICE STEVENS, we do not believe we can interpret the statute's language to save its constitutionality, see post, at 286 (opinion dissenting in part), because we believe that any such reinterpretation, even if limited to instances in which a Sixth Amendment problem arises, would be "plainly contrary to the intent of Congress." United States v. X-Citement Video, Inc., 513 U. S. 64, 78 (1994). Without some such reinterpretation, however, this provision of the statute, along with those inextricably connected to it, are constitutionally invalid, and fall outside of Congress' power to enact. Nor can we agree with JUSTICE STEVENS that a newly passed "identical statute" would be valid, post, at 283 (opinion dissenting in part). Such a new, identically worded statute would be valid only if (unlike the present statute) we could interpret that new statute (without disregarding Congress' basic intent) as being consistent with the Court's jury factfinding requirement. Compare post, at 283-284 (STEVENS, J., dissenting in part). If so, the statute would stand.

Second, Congress' basic statutory goal — a system that diminishes sentencing disparity — depends for its success upon judicial efforts to determine, and to base punishment upon, the real conduct that underlies the crime of conviction. That determination is particularly important in the federal system where crimes defined as, for example, "obstruct[ing], delay[ing], or affect[ing] commerce or the movement of any article or commodity in commerce, by . . . extortion," 18 U. S. C. § 1951(a), or, say, using the mail "for the purpose of executing" a "scheme or artifice to defraud," § 1341 (2000 ed., Supp. II), can encompass a vast range of very different kinds of underlying conduct. But it is also important even in respect to ordinary crimes, such as robbery, where an act that meets the statutory definition can be committed in a host of different ways. Judges have long looked to real conduct when sentencing. Federal judges have long relied upon a presentence report, prepared by a probation officer, for information (often unavailable until after the trial) relevant to the manner in which the convicted offender committed the crime of conviction.

Congress expected this system to continue. That is why it specifically inserted into the Act the provision cited above, which (recodifying prior law) says that

"[n]o limitation shall be placed on the information concerning the background, character, and conduct of a person convicted of an offense which a court of the United States may receive and consider for the purpose of imposing an appropriate sentence." 18 U. S. C. § 3661.

This Court's earlier opinions assumed that this system would continue. That is why the Court, for example, held in United States v. Watts, 519 U. S. 148 (1997) (per curiam), that a sentencing judge could rely for sentencing purposes upon a fact that a jury had found unproved (beyond a reasonable doubt). See id., at 157; see also id., at 152-153 (quoting United States Sentencing Commission, Guidelines Manual § 1B1.3, comment., backg'd (Nov. 1995) (USSG), which "describes in sweeping language the conduct that a sentencing court may consider in determining the applicable guideline range," and which provides that "`[c]onduct that is not formally charged or is not an element of the offense of conviction may enter into the determination of the applicable guideline sentencing range'").

The Sentencing Guidelines also assume that Congress intended this system to continue. See USSG § 1B1.3, comment., backg'd (Nov. 2003). That is why, among other things, they permit a judge to reject a plea-bargained sentence if he determines, after reviewing the presentence report, that the sentence does not adequately reflect the seriousness of the defendant's actual conduct. See § 6B1.2(a).

To engraft the Court's constitutional requirement onto the sentencing statutes, however, would destroy the system. It would prevent a judge from relying upon a presentence report for factual information, relevant to sentencing, uncovered after the trial. In doing so, it would, even compared to pre-Guidelines sentencing, weaken the tie between a sentence and an offender's real conduct. It would thereby undermine the sentencing statute's basic aim of ensuring similar sentences for those who have committed similar crimes in similar ways.

Several examples help illustrate the point. Imagine Smith and Jones, each of whom violates the Hobbs Act in very different ways. See 18 U. S. C. § 1951(a) (forbidding "obstruct[ing], delay[ing], or affect[ing] commerce or the movement of any article or commodity in commerce, by ... extortion"). Smith threatens to injure a co-worker unless the co-worker advances him a few dollars from the interstate company's till; Jones, after similarly threatening the co-worker, causes far more harm by seeking far more money, by making certain that the co-worker's family is aware of the threat, by arranging for deliveries of dead animals to the co-worker's home to show he is serious, and so forth. The offenders' behavior is very different; the known harmful consequences of their actions are different; their punishments both before, and after, the Guidelines would have been different. But, under the dissenters' approach, unless prosecutors decide to charge more than the elements of the crime, the judge would have to impose similar punishments. See, e. g., post, at 303-304 (SCALIA, J., dissenting in part).

Now imagine two former felons, Johnson and Jackson, each of whom engages in identical criminal behavior: threatening a bank teller with a gun, securing $50,000, and injuring an innocent bystander while fleeing the bank. Suppose prosecutors charge Johnson with one crime (say, illegal gun possession, see 18 U. S. C. § 922(g)) and Jackson with another (say, bank robbery, see § 2113(a)). Before the Guidelines, a single judge faced with such similar real conduct would have been able (within statutory limits) to impose similar sentences upon the two similar offenders despite the different charges brought against them. The Guidelines themselves would ordinarily have required judges to sentence the two offenders similarly. But under the dissenters' system, in these circumstances the offenders likely would receive different punishments. See, e. g., post, at 303-304 (SCALIA, J., dissenting in part).

Consider, too, a complex mail fraud conspiracy where a prosecutor may well be uncertain of the amount of harm and of the role each indicted individual played until after conviction — when the offenders may turn over financial records, when it becomes easier to determine who were the leaders and who the followers, when victim interviews are seen to be worth the time. In such a case the relation between the sentence and what actually occurred is likely to be considerably more distant under a system with a jury trial requirement patched onto it than it was even prior to the Sentencing Act, when judges routinely used information obtained after the verdict to decide upon a proper sentence.

This point is critically important. Congress' basic goal in passing the Sentencing Act was to move the sentencing system in the direction of increased uniformity. See 28 U. S. C. § 991(b)(1)(B); see also § 994(f). That uniformity does not consist simply of similar sentences for those convicted of violations of the same statute — a uniformity consistent with the dissenters' remedial approach. It consists, more importantly, of similar relationships between sentences and real conduct, relationships that Congress' sentencing statutes helped to advance and that JUSTICE STEVENS' approach would undermine. Compare post, at 288 (opinion dissenting in part) (conceding that the Sixth Amendment requirement would "undoubtedly affect `real conduct' sentencing in certain cases," but minimizing the significance of that circumstance). In significant part, it is the weakening of this real-conduct/uniformity-in-sentencing relationship, and not any "[i]nexplicabl[e]" concerns for the "manner of achieving uniform sentences," post, at 304 (SCALIA, J., dissenting in part), that leads us to conclude that Congress would have preferred no mandatory system to the system the dissenters envisage.

Third, the sentencing statutes, read to include the Court's Sixth Amendment requirement, would create a system far more complex than Congress could have intended. How would courts and counsel work with an indictment and a jury trial that involved not just whether a defendant robbed a bank but also how? Would the indictment have to allege, in addition to the elements of robbery, whether the defendant possessed a firearm, whether he brandished or discharged it, whether he threatened death, whether he caused bodily injury, whether any such injury was ordinary, serious, permanent or life threatening, whether he abducted or physically restrained anyone, whether any victim was unusually vulnerable, how much money was taken, and whether he was an organizer, leader, manager, or supervisor in a robbery gang? See USSG §§ 2B3.1, 3B1.1. If so, how could a defendant mount a defense against some or all such specific claims should he also try simultaneously to maintain that the Government's evidence failed to place him at the scene of the crime? Would the indictment in a mail fraud case have to allege the number of victims, their vulnerability, and the amount taken from each? How could a judge expect a jury to work with the Guidelines' definitions of, say, "relevant conduct," which includes "all acts and omissions committed, aided, abetted, counseled, commanded, induced, procured, or willfully caused by the defendant; and [in the case of a conspiracy] all reasonably foreseeable acts and omissions of others in furtherance of the jointly undertaken criminal activity"? §§ 1B1.3(a)(1)(A)-(B). How would a jury measure "loss" in a securities fraud case — a matter so complex as to lead the Commission to instruct judges to make "only ... a reasonable estimate"? § 2B1.1, comment., n. 3(C). How would the court take account, for punishment purposes, of a defendant's contemptuous behavior at trial — a matter that the Government could not have charged in the indictment? § 3C1.1.

Fourth, plea bargaining would not significantly diminish the consequences of the Court's constitutional holding for the operation of the Guidelines. Compare post, at 273-274 (STEVENS, J., dissenting in part). Rather, plea bargaining would make matters worse. Congress enacted the sentencing statutes in major part to achieve greater uniformity in sentencing, i. e., to increase the likelihood that offenders who engage in similar real conduct would receive similar sentences. The statutes reasonably assume that their efforts to move the trial-based sentencing process in the direction of greater sentencing uniformity would have a similar positive impact upon plea-bargained sentences, for plea bargaining takes place in the shadow of (i. e., with an eye towards the hypothetical result of) a potential trial.

That, too, is why Congress, understanding the realities of plea bargaining, authorized the Commission to promulgate policy statements that would assist sentencing judges in determining whether to reject a plea agreement after reading about the defendant's real conduct in a presentence report (and giving the offender an opportunity to challenge the report). See 28 U. S. C. § 994(a)(2)(E); USSG § 6B1.2(a), p. s. This system has not worked perfectly; judges have often simply accepted an agreed-upon account of the conduct at issue. But compared to pre-existing law, the statutes try to move the system in the right direction, i. e., toward greater sentencing uniformity.

The Court's constitutional jury trial requirement, however, if patched onto the present Sentencing Act, would move the system backwards in respect both to tried and to plea-bargained cases. In respect to tried cases, it would effectively deprive the judge of the ability to use post-verdict-acquired real-conduct information; it would prohibit the judge from basing a sentence upon any conduct other than the conduct the prosecutor chose to charge; and it would put a defendant to a set of difficult strategic choices as to which prosecutorial claims he would contest. The sentence that would emerge in a case tried under such a system would likely reflect real conduct less completely, less accurately, and less often than did a pre-Guidelines, as well as a Guidelines, trial.

Because plea bargaining inevitably reflects estimates of what would happen at trial, plea bargaining too under such a system would move in the wrong direction. That is to say, in a sentencing system modified by the Court's constitutional requirement, plea bargaining would likely lead to sentences that gave greater weight not to real conduct, but rather to the skill of counsel, the policies of the prosecutor, the case-load, and other factors that vary from place to place, defendant to defendant, and crime to crime. Compared to pre-Guidelines plea bargaining, plea bargaining of this kind would necessarily move federal sentencing in the direction of diminished, not increased, uniformity in sentencing. Compare supra, at 250-252, with post, at 288 (STEVENS, J., dissenting in part). It would tend to defeat, not to further, Congress' basic statutory goal.

Such a system would have particularly troubling consequences with respect to prosecutorial power. Until now, sentencing factors have come before the judge in the presentence report. But in a sentencing system with the Court's constitutional requirement engrafted onto it, any factor that a prosecutor chose not to charge at the plea negotiation would be placed beyond the reach of the judge entirely. Prosecutors would thus exercise a power the Sentencing Act vested in judges: the power to decide, based on relevant information about the offense and the offender, which defendants merit heavier punishment.

In respondent Booker's case, for example, the jury heard evidence that the crime had involved 92.5 grams of crack cocaine, and convicted Booker of possessing more than 50 grams. But the judge, at sentencing, found that the crime had involved an additional 566 grams, for a total of 658.5 grams. A system that would require the jury, not the judge, to make the additional "566 grams" finding is a system in which the prosecutor, not the judge, would control the sentence. That is because it is the prosecutor who would have to decide what drug amount to charge. He could choose to charge 658.5 grams, or 92.5, or less. It is the prosecutor who, through such a charging decision, would control the sentencing range. And it is different prosecutors who, in different cases — say, in two cases involving 566 grams — would potentially insist upon different punishments for similar defendants who engaged in similar criminal conduct involving similar amounts of unlawful drugs — say, by charging one of them with the full 566 grams, and the other with 10. As long as different prosecutors react differently, a system with a patched-on jury factfinding requirement would mean different sentences for otherwise similar conduct, whether in the context of trials or that of plea bargaining.

Fifth, Congress would not have enacted sentencing statutes that make it more difficult to adjust sentences upward than to adjust them downward. As several United States Senators have written in an amicus brief, "the Congress that enacted the 1984 Act did not conceive of — much less establish — a sentencing guidelines system in which sentencing judges were free to consider facts or circumstances not found by a jury or admitted in a plea agreement for the purpose of adjusting a base-offense level down, but not up, within the applicable guidelines range. Such a one-way lever would be grossly at odds with Congress's intent." Brief for Sen. Orrin G. Hatch et al. as Amici Curiae 22. Yet that is the system that the dissenters' remedy would create. Compare post, at 291 (STEVENS, J., dissenting in part) (conceding asymmetry but stating belief that this "is unlikely to have more than a minimal effect").

For all these reasons, Congress, had it been faced with the constitutional jury trial requirement, likely would not have passed the same Sentencing Act. It likely would have found the requirement incompatible with the Act as written. Hence the Act cannot remain valid in its entirety. Severance and excision are necessary.

III

We now turn to the question of which portions of the sentencing statute we must sever and excise as inconsistent with the Court's constitutional requirement. Although, as we have explained, see Part II, supra, we believe that Congress would have preferred the total invalidation of the statute to the dissenters' remedial approach, we nevertheless do not believe that the entire statute must be invalidated. Compare post, at 292 (STEVENS, J., dissenting in part). Most of the statute is perfectly valid. See, e.g., 18 U. S. C. A. § 3551 (main ed. and Supp. 2004) (describing authorized sentences as probation, fine, or imprisonment); § 3552 (presentence reports); § 3554 (forfeiture); § 3555 (notification to the victims); § 3583 (supervised release). And we must "refrain from invalidating more of the statute than is necessary." Regan, 468 U.S., at 652 (plurality opinion). Indeed, we must retain those portions of the Act that are (1) constitutionally valid, id., at 652-653, (2) capable of "functioning independently," Alaska Airlines, 480 U. S., at 684, and (3) consistent with Congress' basic objectives in enacting the statute, Regan, supra, at 653.

Application of these criteria indicates that we must sever and excise two specific statutory provisions: the provision that requires sentencing courts to impose a sentence within the applicable Guidelines range (in the absence of circumstances that justify a departure), see 18 U. S. C. A. § 3553(b)(1) (Supp. 2004), and the provision that sets forth standards of review on appeal, including de novo review of departures from the applicable Guidelines range, see § 3742(e) (main ed. and Supp. 2004) (see Appendix, infra, for text of both provisions). With these two sections excised (and statutory cross-references to the two sections consequently invalidated), the remainder of the Act satisfies the Court's constitutional requirements.

As the Court today recognizes in its first opinion in these cases, the existence of § 3553(b)(1) is a necessary condition of the constitutional violation. That is to say, without this provision — namely the provision that makes "the relevant sentencing rules . . . mandatory and impose[s] binding requirements on all sentencing judges" — the statute falls outside the scope of Apprendi's requirement. Ante, at 233 (STEVENS, J., opinion of the Court); see also ibid. ("[E]veryone agrees that the constitutional issues presented by these cases would have been avoided entirely if Congress had omitted from the [Sentencing Reform Act] the provisions that make the Guidelines binding on district judges"). Cf. post, at 314-320 (THOMAS, J., dissenting in part).

The remainder of the Act "function[s] independently." Alaska Airlines, supra, at 684. Without the "mandatory" provision, the Act nonetheless requires judges to take account of the Guidelines together with other sentencing goals. See 18 U. S. C. A. § 3553(a) (Supp. 2004). The Act nonetheless requires judges to consider the Guidelines "sentencing range established for . . . the applicable category of offense committed by the applicable category of defendant," § 3553(a)(4)(A), the pertinent Sentencing Commission policy statements, the need to avoid unwarranted sentencing disparities, and the need to provide restitution to victims, §§ 3553(a)(1), (3), (5)-(7) (main ed. and Supp. 2004). And the Act nonetheless requires judges to impose sentences that reflect the seriousness of the offense, promote respect for the law, provide just punishment, afford adequate deterrence, protect the public, and effectively provide the defendant with needed educational or vocational training and medical care. § 3553(a)(2) (main ed. and Supp. 2004) (see Appendix, infra, for text of § 3553(a)).

Moreover, despite the absence of § 3553(b)(1) (Supp. 2004), the Act continues to provide for appeals from sentencing decisions (irrespective of whether the trial judge sentences within or outside the Guidelines range in the exercise of his discretionary power under § 3553(a)). See § 3742(a) (appeal by defendant); § 3742(b) (appeal by Government). We concede that the excision of § 3553(b)(1) requires the excision of a different, appeals-related section, namely § 3742(e) (main ed. and Supp. 2004), which sets forth standards of review on appeal. That section contains critical cross-references to the (now-excised) § 3553(b)(1) and consequently must be severed and excised for similar reasons.

Excision of § 3742(e), however, does not pose a critical problem for the handling of appeals. That is because, as we have previously held, a statute that does not explicitly set forth a standard of review may nonetheless do so implicitly. See Pierce v. Underwood, 487 U. S. 552, 558-560 (1988) (adopting a standard of review, where "neither a clear statutory prescription nor a historical tradition" existed, based on the statutory text and structure, and on practical considerations); see also Cooter & Gell v. Hartmarx Corp., 496 U. S. 384, 403-405 (1990) (same); Koon v. United States, 518 U. S. 81, 99 (1996) (citing Pierce and Cooter & Gell with approval). We infer appropriate review standards from related statutory language, the structure of the statute, and the "`sound administration of justice.'" Pierce, supra, at 559-560. And in this instance those factors, in addition to the past two decades of appellate practice in cases involving departures, imply a practical standard of review already familiar to appellate courts: review for "unreasonable[ness]." 18 U. S. C. § 3742(e)(3) (1994 ed.).

Until 2003, § 3742(e) explicitly set forth that standard. See § 3742(e)(3) (1994 ed.). In 2003, Congress modified the pre-existing text, adding a de novo standard of review for departures and inserting cross-references to § 3553(b)(1). Prosecutorial Remedies and Other Tools to end the Exploitation of Children Today Act of 2003, Pub. L. 108-21, § 401(d)(1), 117 Stat. 670. In light of today's holding, the reasons for these revisions — to make Guidelines sentencing even more mandatory than it had been — have ceased to be relevant. The pre-2003 text directed appellate courts to review sentences that reflected an applicable Guidelines range for correctness, but to review other sentences — those that fell "outside the applicable Guideline range" — with a view toward determining whether such a sentence

"is unreasonable, having regard for ... the factors to be considered in imposing a sentence, as set forth in chapter 227 of this title; and . . . the reasons for the imposition of the particular sentence, as stated by the district court pursuant to the provisions of section 3553(c)." 18 U. S. C. § 3742(e)(3) (1994 ed.) (emphasis added).

In other words, the text told appellate courts to determine whether the sentence "is unreasonable" with regard to § 3553(a). Section 3553(a) remains in effect, and sets forth numerous factors that guide sentencing. Those factors in turn will guide appellate courts, as they have in the past, in determining whether a sentence is unreasonable.

Taking into account the factors set forth in Pierce, we read the statute as implying this appellate review standard — a standard consistent with appellate sentencing practice during the last two decades. JUSTICE SCALIA believes that only in "Wonderland" is it possible to infer a standard of review after excising § 3742(e). See post, at 309 (opinion dissenting in part). But our application of Pierce does not justify that characterization. Pierce requires us to judge the appropriateness of our inference based on the statute's language and basic purposes. We believe our inference a fair one linguistically, and one consistent with Congress' intent to provide appellate review. Under these circumstances, to refuse to apply Pierce and thereby retreat to a remedy that raises the problems discussed in Part II, supra (as the dissenters would do), or thereby eliminate appellate review entirely, would cut the statute loose from its moorings in congressional purpose.

Nor do we share the dissenters' doubts about the practicality of a "reasonableness" standard of review. "Reasonableness" standards are not foreign to sentencing law. The Act has long required their use in important sentencing circumstances — both on review of departures, see 18 U. S. C. § 3742(e)(3) (1994 ed.), and on review of sentences imposed where there was no applicable Guideline, see §§ 3742(a)(4), (b)(4), (e)(4). Together, these cases account for about 16.7% of sentencing appeals. See United States Sentencing Commission, 2002 Sourcebook of Federal Sentencing Statistics 107, n. 1, 111 (at least 711 of 5,018 sentencing appeals involved departures), 108 (at least 126 of 5,018 sentencing appeals involved the imposition of a term of imprisonment after the revocation of supervised release). See also, e. g., United States v. White Face, 383 F. 3d 733, 737-740 (CA8 2004); United States v. Tsosie, 376 F. 3d 1210, 1218-1219 (CA10 2004); United States v. Salinas, 365 F. 3d 582, 588-590 (CA7 2004); United States v. Cook, 291 F. 3d 1297, 1300-1302 (CA11 2002 (per curiam); United States v. Olabanji, 268 F. 3d 636, 637-639 (CA9 2001); United States v. Ramirez-Rivera, 241 F. 3d 37, 40-41 (CA1 2001). That is why we think it fair (and not, in JUSTICE SCALIA's words, a "gross exaggeratio[n]," post, at 311 (opinion dissenting in part)), to assume judicial familiarity with a "reasonableness" standard. And that is why we believe that appellate judges will prove capable of facing with greater equanimity than would JUSTICE SCALIA what he calls the "daunting prospect," post, at 312, of applying such a standard across the board.

Neither do we share JUSTICE SCALIA's belief that use of a reasonableness standard "will produce a discordant symphony" leading to "excessive sentencing disparities," and "wreak havoc" on the judicial system, post, at 312-313 (internal quotation marks omitted). The Sentencing Commission will continue to collect and study appellate court decision-making. It will continue to modify its Guidelines in light of what it learns, thereby encouraging what it finds to be better sentencing practices. It will thereby promote uniformity in the sentencing process. 28 U. S. C. A. § 994 (main ed. and Supp. 2004).

Regardless, in this context, we must view fears of a "discordant symphony," "excessive disparities," and "havoc" (if they are not themselves "gross exaggerations") with a comparative eye. We cannot and do not claim that use of a "reasonableness" standard will provide the uniformity that Congress originally sought to secure. Nor do we doubt that Congress wrote the language of the appellate provisions to correspond with the mandatory system it intended to create. Compare post, at 306-307 (SCALIA, J., dissenting in part) (expressing concern regarding the presence of § 3742(f) in light of the absence of § 3742(e)). But, as by now should be clear, that mandatory system is no longer an open choice. And the remedial question we must ask here (as we did in respect to § 3553(b)(1)) is, which alternative adheres more closely to Congress' original objective: (1) retention of sentencing appeals, or (2) invalidation of the entire Act, including its appellate provisions? The former, by providing appellate review, would tend to iron out sentencing differences; the latter would not. Hence we believe Congress would have preferred the former to the latter — even if the former means that some provisions will apply differently from the way Congress had originally expected. See post, at 306-307 (SCALIA, J., dissenting in part). But, as we have said, we believe that Congress would have preferred even the latter to the system the dissenters recommend, a system that has its own problems of practicality. See supra, at 254-256.

Finally, the Act without its "mandatory" provision and related language remains consistent with Congress' initial and basic sentencing intent. Congress sought to "provide certainty and fairness in meeting the purposes of sentencing, [while] avoiding unwarranted sentencing disparities ... [and] maintaining sufficient flexibility to permit individualized sentences when warranted." 28 U.S.C. § 991(b)(1)(B); see also USSG § 1A1.1, application note (explaining that Congress sought to achieve "honesty," "uniformity," and "proportionality" in sentencing (emphases deleted)). The system remaining after excision, while lacking the mandatory features that Congress enacted, retains other features that help to further these objectives.

As we have said, the Sentencing Commission remains in place, writing Guidelines, collecting information about actual district court sentencing decisions, undertaking research, and revising the Guidelines accordingly. See 28 U. S. C. A. § 994 (main ed. and Supp. 2004). The district courts, while not bound to apply the Guidelines, must consult those Guidelines and take them into account when sentencing. See 18 U. S. C. A. §§ 3553(a)(4), (5)(Supp. 2004). But compare post, at 305 (SCALIA, J., dissenting in part) (claiming that the sentencing judge has the same discretion "he possessed before the Act was passed"). The courts of appeals review sentencing decisions for unreasonableness. These features of the remaining system, while not the system Congress enacted, nonetheless continue to move sentencing in Congress' preferred direction, helping to avoid excessive sentencing disparities while maintaining flexibility sufficient to individualize sentences where necessary. See 28 U. S. C. § 991(b). We can find no feature of the remaining system that tends to hinder, rather than to further, these basic objectives. Under these circumstances, why would Congress not have preferred excision of the "mandatory" provision to a system that engrafts today's constitutional requirement onto the unchanged pre-existing statute — a system that, in terms of Congress' basic objectives, is counterproductive?

We do not doubt that Congress, when it wrote the Sentencing Act, intended to create a form of mandatory Guidelines system. See post, at 291-296 (STEVENS, J., dissenting in part). But, we repeat, given today's constitutional holding, that is not a choice that remains open. Hence we have examined the statute in depth to determine Congress' likely intent in light of today's holding. See, e. g., Denver Area Ed. Telecommunications Consortium, Inc., 518 U. S., at 767. And we have concluded that today's holding is fundamentally inconsistent with the judge-based sentencing system that Congress enacted into law. In our view, it is more consistent with Congress' likely intent in enacting the Sentencing Reform Act (1) to preserve important elements of that system while severing and excising two provisions (§§ 3553(b)(1) and 3742(e)) than (2) to maintain all provisions of the Act and engraft today's constitutional requirement onto that statutory scheme.

Ours, of course, is not the last word: The ball now lies in Congress' court. The National Legislature is equipped to devise and install, long term, the sentencing system, compatible with the Constitution, that Congress judges best for the federal system of justice.

IV

We briefly explain why we have not fully adopted the remedial proposals that the parties have advanced. First, the Government argues that "in any case in which the Constitution prohibits the judicial factfinding procedures that Congress and the Commission contemplated for implementing the Guidelines, the Guidelines as a whole become inapplicable." Brief for United States in No. 04-104, p. 44. Thus the Guidelines "system contemplated by Congress and created by the Commission would be inapplicable in a case in which the Guidelines would require the sentencing court to find a sentence-enhancing fact." Id., at 66-67. The Guidelines would remain advisory, however, for § 3553(a) would remain intact. Ibid. Cf. Brief for New York Council of Defense Lawyers as Amicus Curiae 15, n. 9 (A "decision that Section 3553(b) ... is unconstitutional ... would not necessarily jeopardize the other reforms made by the Sentencing Reform Act, including ... 18 U. S. C. § 3553(a)"); see also ibid. (recognizing that the remainder of the Act functions independently); Brief for Families Against Mandatory Minimums as Amicus Curiae 29, 30.

As we understand the Government's remedial suggestion, it coincides significantly with our own. But compare post, at 282 (STEVENS, J., dissenting in part) (asserting that no party or amicus sought the remedy we adopt); post, at 309 (SCALIA, J., dissenting in part) (same). The Government would render the Guidelines advisory in "any case in which the Constitution prohibits" judicial factfinding. But it apparently would leave them as binding in all other cases.

We agree with the first part of the Government's suggestion. However, we do not see how it is possible to leave the Guidelines as binding in other cases. For one thing, the Government's proposal would impose mandatory Guidelines-type limits upon a judge's ability to reduce sentences, but it would not impose those limits upon a judge's ability to increase sentences. We do not believe that such "one-way lever[s]" are compatible with Congress' intent. Cf. Brief for Sen. Orrin G. Hatch et al. as Amici Curiae 22; see also supra, at 253-254. For another, we believe that Congress would not have authorized a mandatory system in some cases and a nonmandatory system in others, given the administrative complexities that such a system would create. Such a two-system proposal seems unlikely to further Congress' basic objective of promoting uniformity in sentencing.

Second, the respondents in essence would take the same approach as would JUSTICE STEVENS. They believe that the constitutional requirement is compatible with the Sentencing Act, and they ask us to hold that the Act continues to stand as written with the constitutional requirement engrafted onto it. We do not accept their position for the reasons we have already given. See Part II, supra.

Respondent Fanfan argues in the alternative that we should excise those provisions of the Sentencing Act that require judicial factfinding at sentencing. That system, however, would produce problems similar to those we have discussed in Part II, supra. We reject Fanfan's remedial suggestion for that reason.

V

In respondent Booker's case, the District Court applied the Guidelines as written and imposed a sentence higher than the maximum authorized solely by the jury's verdict. The Court of Appeals held Blakely applicable to the Guidelines, concluded that Booker's sentence violated the Sixth Amendment, vacated the judgment of the District Court, and remanded for resentencing. We affirm the judgment of the Court of Appeals and remand the case. On remand, the District Court should impose a sentence in accordance with today's opinions, and, if the sentence comes before the Court of Appeals for review, the Court of Appeals should apply the review standards set forth in this opinion.

In respondent Fanfan's case, the District Court held Blakely applicable to the Guidelines. It then imposed a sentence that was authorized by the jury's verdict — a sentence lower than the sentence authorized by the Guidelines as written. Thus, Fanfan's sentence does not violate the Sixth Amendment. Nonetheless, the Government (and the defendant should he so choose) may seek resentencing under the system set forth in today's opinions. Hence we vacate the judgment of the District Court and remand the case for further proceedings consistent with this opinion.

As these dispositions indicate, we must apply today's holdings — both the Sixth Amendment holding and our remedial interpretation of the Sentencing Act — to all cases on direct review. See Griffith v. Kentucky, 479 U. S. 314, 328 (1987) ("[A] new rule for the conduct of criminal prosecutions is to be applied retroactively to all cases ... pending on direct review or not yet final, with no exception for cases in which the new rule constitutes a `clear break' with the past"). See also Reynoldsville Casket Co. v. Hyde, 514 U. S. 749, 752 (1995) (civil case); Harper v. Virginia Dept. of Taxation, 509 U. S. 86, 97 (1993) (same). That fact does not mean that we believe that every sentence gives rise to a Sixth Amendment violation. Nor do we believe that every appeal will lead to a new sentencing hearing. That is because we expect reviewing courts to apply ordinary prudential doctrines, determining, for example, whether the issue was raised below and whether it fails the "plain-error" test. It is also because, in cases not involving a Sixth Amendment violation, whether resentencing is warranted or whether it will instead be sufficient to review a sentence for reasonableness may depend upon application of the harmless-error doctrine.

It is so ordered.

APPENDIX TO OPINION OF THE COURT

Title 18 U. S. C. A. § 3553(a) (main ed. and Supp. 2004) provides:

"Factors to be considered in imposing a sentence. — The court shall impose a sentence sufficient, but not greater than necessary, to comply with the purposes set forth in paragraph (2) of this subsection. The court, in determining the particular sentence to be imposed, shall consider —

"(1) the nature and circumstances of the offense and the history and characteristics of the defendant;

"(2) the need for the sentence imposed —

"(A) to reflect the seriousness of the offense, to promote respect for the law, and to provide just punishment for the offense;

"(B) to afford adequate deterrence to criminal conduct;

"(C) to protect the public from further crimes of the defendant; and

"(D) to provide the defendant with needed educational or vocational training, medical care, or other correctional treatment in the most effective manner;

"(3) the kinds of sentences available;

"(4) the kinds of sentence and the sentencing range established for —

"(A) the applicable category of offense committed by the applicable category of defendant as set forth in the guidelines —

"(i) issued by the Sentencing Commission pursuant to section 994(a)(1) of title 28, United States Code, subject to any amendments made to such guidelines by act of Congress (regardless of whether such amendments have yet to be incorporated by the Sentencing Commission into amendments issued under section 994(p) of title 28); and

"(ii) that, except as provided in section 3742(g), are in effect on the date the defendant is sentenced; or

"(B) in the case of a violation of probation or supervised release, the applicable guidelines or policy statements issued by the Sentencing Commission pursuant to section 994(a)(3) of title 28, United States Code, taking into account any amendments made to such guidelines or policy statements by act of Congress (regardless of whether such amendments have yet to be incorporated by the Sentencing Commission into amendments issued under section 994(p) of title 28);

"(5) any pertinent policy statement —

"(A) issued by the Sentencing Commission pursuant to section 994(a)(2) of title 28, United States Code, subject to any amendments made to such policy statement by act of Congress (regardless of whether such amendments have yet to be incorporated by the Sentencing Commission into amendments issued under section 994(p) of title 28); and

"(B) that, except as provided in section 3742(g), is in effect on the date the defendant is sentenced.

"(6) the need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct; and

"(7) the need to provide restitution to any victims of the offense."

Title 18 U. S. C. A. § 3553(b)(1) (Supp. 2004) provides: "Application of guidelines in imposing a sentence. — (1) In general. — Except as provided in paragraph (2), the court shall impose a sentence of the kind, and within the range, referred to in subsection (a)(4) unless the court finds that there exists an aggravating or mitigating circumstance of a kind, or to a degree, not adequately taken into consideration by the Sentencing Commission in formulating the guidelines that should result in a sentence different from that described. In determining whether a circumstance was adequately taken into consideration, the court shall consider only the sentencing guidelines, policy statements, and official commentary of the Sentencing Commission. In the absence of an applicable sentencing guideline, the court shall impose an appropriate sentence, having due regard for the purposes set forth in subsection (a)(2). In the absence of an applicable sentencing guideline in the case of an offense other than a petty offense, the court shall also have due regard for the relationship of the sentence imposed to sentences prescribed by guidelines applicable to similar offenses and offenders, and to the applicable policy statements of the Sentencing Commission."

Title 18 U. S. C. A. § 3742(e) (main ed. and Supp. 2004) provides:

"Consideration. — Upon review of the record, the court of appeals shall determine whether the sentence —

"(1) was imposed in violation of law;

"(2) was imposed as a result of an incorrect application of the sentencing guidelines;

"(3) is outside the applicable guideline range, and

"(A) the district court failed to provide the written statement of reasons required by section 3553(c);

"(B) the sentence departs from the applicable guideline range based on a factor that —

"(i) does not advance the objectives set forth in section 3553(a)(2); or

"(ii) is not authorized under section 3553(b); or

"(iii) is not justified by the facts of the case; or

"(C) the sentence departs to an unreasonable degree from the applicable guidelines range, having regard for the factors to be considered in imposing a sentence, as set forth in section 3553(a) of this title and the reasons for the imposition of the particular sentence, as stated by the district court pursuant to the provisions of section 3553(c); or

"(4) was imposed for an offense for which there is no applicable sentencing guideline and is plainly unreasonable.

"The court of appeals shall give due regard to the opportunity of the district court to judge the credibility of the witnesses, and shall accept the findings of fact of the district court unless they are clearly erroneous and, except with respect to determinations under subsection (3)(A) or (3)(B), shall give due deference to the district court's application of the guidelines to the facts. With respect to determinations under subsection (3)(A) or (3)(B), the court of appeals shall review de novo the district court's application of the guidelines to the facts."

JUSTICE STEVENS, with whom JUSTICE SOUTER joins, and with whom JUSTICE SCALIA joins except for Part III and footnote 17, dissenting in part.

Neither of the two Court opinions that decide these cases finds any constitutional infirmity inherent in any provision of the Sentencing Reform Act of 1984 (SRA) or the Federal Sentencing Guidelines. Specifically, neither 18 U. S. C. A. § 3553(b)(1) (Supp. 2004), which makes application of the Guidelines mandatory, nor § 3742(e) (main ed. and Supp. 2004), which authorizes appellate review of departures from the Guidelines, is even arguably unconstitutional. Neither the Government, nor the respondents, nor any of the numerous amici has suggested that there is any need to invalidate either provision in order to avoid violations of the Sixth Amendment in the administration of the Guidelines. The Court's decision to do so represents a policy choice that Congress has considered and decisively rejected. While it is perfectly clear that Congress has ample power to repeal these two statutory provisions if it so desires, this Court should not make that choice on Congress' behalf. I respectfully dissent from the Court's extraordinary exercise of authority.

Before explaining why the law does not authorize the Court's creative remedy, why the reasons it advances in support of its decision are unpersuasive, and why it is abundantly clear that Congress has already rejected that very remedy, it is appropriate to explain how the violation of the Sixth Amendment that occurred in Booker's case could readily have been avoided without making any change in the Guidelines. Booker received a sentence of 360 months' imprisonment. His sentence was based on four factual determinations: (1) the jury's finding that he possessed 92.5 grams of crack (cocaine base); (2) the judge's finding that he possessed an additional 566 grams; (3) the judge's conclusion that he had obstructed justice; and (4) the judge's evaluation of his prior criminal record. Under the jury's 92.5 grams finding, the maximum sentence authorized by the Guidelines was a term of 262 months. See United States Sentencing Commission, Guidelines Manual § 2D1.1(c)(4) (Nov. 2003) (USSG).

If the 566 gram finding had been made by the jury based on proof beyond a reasonable doubt, that finding would have authorized a Guidelines sentence anywhere between 324 and 405 months — the equivalent of a range from 27 to nearly 34 years — given Booker's criminal history. § 2D1.1(c)(2). Relying on his own appraisal of the defendant's obstruction of justice, and presumably any other information in the presentence report, the judge would have had discretion to select any sentence within that range. Thus, if the two facts, which in this case actually established two separate crimes, had both been found by the jury, the judicial factfinding that produced the actual sentence would not have violated the Constitution. In other words, the judge could have considered Booker's obstruction of justice, his criminal history, and all other real offense and offender factors without violating the Sixth Amendment. Because the Guidelines as written possess the virtue of combining a mandatory determination of sentencing ranges and discretionary decisions within those ranges, they allow ample latitude for judicial factfinding that does not even arguably raise any Sixth Amendment issue.

The principal basis for the Court's chosen remedy is its assumption that Congress did not contemplate that the Sixth Amendment would be violated by depriving the defendant of the right to a jury trial on a factual issue as important as whether Booker possessed the additional 566 grams of crack that exponentially increased the maximum sentence that he could receive. I am not at all sure that that assumption is correct, but even if it is, it does not provide an adequate basis for volunteering a systemwide remedy that Congress has already rejected and could enact on its own if it elected to.

When one pauses to note that over 95% of all federal criminal prosecutions are terminated by a plea bargain, and the further fact that in almost half of the cases that go to trial there are no sentencing enhancements, the extraordinary overbreadth of the Court's unprecedented remedy is manifest. It is, moreover, unique because, under the Court's reasoning, if Congress should decide to reenact the exact text of the two provisions that the Court has chosen to invalidate, that reenactment would be unquestionably constitutional. In my judgment, it is therefore clear that the Court's creative remedy is an exercise of legislative, rather than judicial, power.

I

It is a fundamental premise of judicial review that all Acts of Congress are presumptively valid. See Regan v. Time, Inc., 468 U. S. 641, 652 (1984). "A ruling of unconstitutionality frustrates the intent of the elected representatives of the people." Ibid. In the past, because of its respect for the coordinate branches of Government, the Court has invalidated duly enacted statutes — or particular provisions of such statutes — "only upon a plain showing that Congress has exceeded its constitutional bounds." United States v. Morrison, 529 U. S. 598, 607 (2000); see also El Paso & Northeastern R. Co. v. Gutierrez, 215 U. S. 87, 97 (1909). The exercise of such power is traditionally limited to issues presented in the case or controversy before the Court, and to the imposition of remedies that redress specific constitutional violations.

There are two narrow exceptions to this general rule. A facial challenge may succeed if a legislative scheme is unconstitutional in all or nearly all of its applications. That is certainly not true in these cases, however, because most applications of the Guidelines are unquestionably valid. A second exception involves cases in which an invalid provision or application cannot be severed from the remainder of the statute. That exception is inapplicable because there is no statutory or Guidelines provision that is invalid. Neither exception supports the majority's newly minted remedy.

Facial Invalidity:

Regardless of how the Court defines the standard for determining when a facial challenge to a statute should succeed,^[1] it is abundantly clear that the fact that a statute, or any provision of a statute, is unconstitutional in a portion of its applications does not render the statute or provision invalid, and no party suggests otherwise. The Government conceded at oral argument that 45% of federal sentences involve no enhancements. Cf. United States Sentencing Commission, 2002 Sourcebook of Federal Sentencing Statistics 39-40 (hereinafter Sourcebook).^[2] And, according to two U. S. Sentencing Commissioners who testified before Congress shortly after we handed down our decision in Blakely v. Washington, 542 U. S. 296 (2004), the number of enhancements that would actually implicate a defendant's Sixth Amendment rights is even smaller. See Hearings on Blakely v. Washington and the Future of the Federal Sentencing Guidelines before the Senate Committee on the Judiciary, 108th Cong., 2d Sess., 2 (2004) (hereinafter Hearings on Blakely) (testimony of Commissioners John R. Steer and Hon. William K. Sessions III) ("[A] majority of the cases sentenced under the federal guidelines do not receive sentencing enhancements that could potentially implicate Blakely"), available at http://www.ussc.gov/hearings/BlakelyTest.pdf (all Internet materials as visited Jan. 7, 2005, and available in Clerk of Court's case file). Simply stated, the Government's submissions to this Court and to Congress demonstrate that the Guidelines could be constitutionally applied in their entirety, without any modifications, in the "majority of the cases sentenced under the federal guidelines." Ibid. On the basis of these submissions alone, this Court should have declined to find the Guidelines, or any particular provisions of the Guidelines, facially invalid.^[3]

Accordingly, the majority's claim that a jury factfinding requirement would "destroy the system," ante, at 252 (opinion of BREYER, J.), would at most apply to a minority of sentences imposed under the Guidelines. In reality, given that the Government and judges have been apprised of the requirements of the Sixth Amendment, the number of unconstitutional applications would have been even smaller had we allowed them the opportunity to comply with our constitutional holding. This is so for several reasons.

First, it is axiomatic that a defendant may waive his Sixth Amendment right to trial by jury. Patton v. United States, 281 U. S. 276, 312-313 (1930). In Blakely we explained that "[w]hen a defendant pleads guilty, the State is free to seek judicial sentence enhancements so long as the defendant either stipulates to the relevant facts or consents to judicial factfinding." 542 U. S., at 310. Such reasoning applies with equal force to sentences imposed under the Guidelines. As the majority concedes, ante, at 248, only a tiny fraction of federal prosecutions ever go to trial. See Estimate, at 2 ("In FY02, 97.1 percent of cases sentenced under the guidelines were the result of plea agreements"). If such procedures were followed in the future, our holding that Blakely applies to the Guidelines would be consequential only in the tiny portion of prospective sentencing decisions that are made after a defendant has been found guilty by a jury.

Second, in the remaining fraction of cases that result in a jury trial, I am confident that those charged with complying with the Guidelines — judges, aided by prosecutors and defense attorneys — could adequately protect defendants' Sixth Amendment rights without this Court's extraordinary remedy. In many cases, prosecutors could avoid an Apprendi v. New Jersey, 530 U. S. 466 (2000), problem simply by alleging in the indictment the facts necessary to reach the chosen Guidelines sentence. Following our decision in Apprendi, and again after our decision in Blakely, the Department of Justice advised federal prosecutors to adopt practices that would enable them "to charge and prove to the jury facts that increase the statutory maximum — for example, drug type and quantity for offenses under 21 U. S. C. 841."^[4] Enhancing the specificity of indictments would be a simple matter, for example, in prosecutions under the federal drug statutes (such as Booker's prosecution). The Government has already directed its prosecutors to allege facts such as the possession of a dangerous weapon or "that the defendant was an organizer or leader of criminal activity that involved five or more participants" in the indictment and prove them to the jury beyond a reasonable doubt.^[5]

Third, even in those trials in which the Guidelines require the finding of facts not alleged in the indictment, such factfinding by a judge is not unconstitutional per se. To be clear, our holding in Parts I-III, ante, at 243-244 (STEVENS, J., opinion of the Court), that Blakely applies to the Guidelines does not establish the "impermissibility of judicial factfinding." Brief for United States 46. Instead, judicial factfinding to support an offense level determination or an enhancement is only unconstitutional when that finding raises the sentence beyond the sentence that could have lawfully been imposed by reference to facts found by the jury or admitted by the defendant. This distinction is crucial to a proper understanding of why the Guidelines could easily function as they are currently written.

Consider, for instance, a case in which the defendant's initial sentencing range under the Guidelines is 130-to-162 months, calculated by combining a base offense level of 28 and a criminal history category of V. See USSG ch. 5, pt. A (Table). Depending upon the particular offense, the sentencing judge may use her discretion to select any sentence within this range, even if her selection relies upon factual determinations beyond the facts found by the jury. If the defendant described above also possessed a firearm, the Guidelines would direct the judge to apply a two-level enhancement under § 2D1.1, which would raise the defendant's total offense level from 28 to 30. That, in turn, would raise the defendant's eligible sentencing range to 151-to-188 months. That act of judicial factfinding would comply with the Guidelines and the Sixth Amendment so long as the sentencing judge then selected a sentence between 151-to-162 months — the lower number (151) being the bottom of offense level 30 and the higher number (162) being the maximum sentence under level 28, which is the upper limit of the range supported by the jury findings alone. This type of overlap between sentencing ranges is the rule, not the exception, in the Guidelines as currently constituted. See 1 Practice Under the Federal Sentencing Guidelines § 6.01[B], p. 7 (P. Bamberger & D. Gottlieb eds. 4th ed. 2003 Supp.) (noting that nearly all Guidelines ranges overlap and that "because of the overlap, the actual sentence imposed can theoretically be the same no matter which guideline range is chosen"). Trial courts have developed considerable expertise in employing overlapping provisions in such a manner as to avoid unnecessary resolution of factual disputes, see § 7.03[B][2], at 34 (2004 Supp.), and lower courts have shown themselves capable of distinguishing proper from improper applications of sentencing enhancements under Blakely, see, e. g., United States v. Mayfield, 386 F.3d 1301 (CA9 2004) (upholding a two-level enhancement for firearm possession from offense level 34 to 36 because the sentencing judge selected a sentence within the overlapping range between the two levels). The interaction of these various Guidelines provisions demonstrates the fallacy in the assumption that judicial factfinding can never be constitutional under the Guidelines.

The majority's answer to the fact that the vast majority of applications of the Guidelines are constitutional is that "we must determine likely intent not by counting proceedings, but by evaluating the consequences of the Court's constitutional requirement" on every imaginable case. Ante, at 248 (opinion of BREYER, J.). That approach ignores the lessons of our facial invalidity cases. Those cases stress that this Court is ill suited to the task of drafting legislation and that, therefore, as a matter of respect for coordinate branches of Government, we ought to presume whenever possible that those charged with writing and implementing legislation will and can apply "the statute consistently with the constitutional command." Time, Inc. v. Hill, 385 U.S. 374, 397 (1967). Indeed, this Court has generally refused to consider "every conceivable situation which might possibly arise in the application of complex and comprehensive legislation," Barrows v. Jackson, 346 U.S. 249, 256 (1953), because "[t]he delicate power of pronouncing an Act of Congress unconstitutional is not to be exercised with reference to hypothetical cases thus imagined," United States v. Raines, 362 U.S. 17, 22 (1960). The Government has already shown it can apply the Guidelines constitutionally even as written, and Congress is perfectly capable of redrafting the statute on its own. Thus, there is no justification for the extreme judicial remedy of total invalidation of any part of the SRA or the Guidelines.

In sum, it is indisputable that the vast majority of federal sentences under the Guidelines would have complied with the Sixth Amendment without the Court's extraordinary remedy. Under any reasonable reading of our precedents, in no way can it be said that the Guidelines are, or that any particular Guidelines provision is, facially unconstitutional.

Severability:

Even though a statute is not facially invalid, a holding that certain specific provisions are unconstitutional may make it necessary to invalidate the entire statute. See generally Stern, Separability and Separability Clauses in the Supreme Court, 51 Harv. L. Rev. 76 (1937) (hereinafter Stern). Our normal rule, however, is that the "unconstitutionality of a part of an Act does not necessarily defeat or affect the validity of its remaining provisions. Unless it is evident that the legislature would not have enacted those provisions which are within its power, independently of that which is not, the invalid part may be dropped if what is left is fully operative as a law." Champlin Refining Co. v. Corporation Comm'n of Okla., 286 U. S. 210, 234 (1932) (emphasis added).^[6]

Our "severability" precedents, however, cannot support the Court's remedy because there is no provision of the SRA or the Guidelines that falls outside of Congress' power. See Alaska Airlines, Inc. v. Brock, 480 U.S. 678, 684 (1987). Accordingly, severability analysis simply does not apply.

The majority concludes that our constitutional holding requires the invalidation of §§ 3553(b)(1) and 3742(e). The first of these sections uses the word "shall" to make the substantive provisions of the Guidelines mandatory. See Mistretta v. United States, 488 U.S. 361, 367 (1989). The second authorizes de novo review of sentencing judges' applications of relevant Guidelines provisions. Neither section is unconstitutional. While these provisions can in certain cases, when combined with other statutory and Guidelines provisions, result in a violation of the Sixth Amendment, they are plainly constitutional on their faces.

Rather than rely on traditional principles of facial invalidity or severability, the majority creates a new category of cases in which this Court may invalidate any part or parts of a statute (and add others) when it concludes that Congress would have preferred a modified system to administering the statute in compliance with the Constitution. This is entirely new law. Usually the Court first declares unconstitutional a particular provision of law, and only then does it inquire whether the remainder of the statute can be saved. See, e.g., Regan v. Time, 468 U.S., at 652; Alaska Airlines, 480 U. S., at 684. Review in this manner limits judicial power by minimizing the damage done to the statute by judicial fiat. There is no case of which I am aware, however, in which this Court has used "severability" analysis to do what the majority does today: determine that some unconstitutional applications of a statute, when viewed in light of the Court's reading of "likely" legislative intent, justifies the invalidation of certain statutory sections in their entirety, their constitutionality notwithstanding, in order to save the parts of the statute the Court deemed most important. The novelty of this remedial maneuver perhaps explains why no party or amicus curiae to this litigation has requested the remedy the Court now orders. In addition, none of the federal courts that have addressed Blakely's application to the Guidelines has concluded that striking down § 3553(b)(1) is a proper solution.

Most importantly, the Court simply has no authority to invalidate legislation absent a showing that it is unconstitutional. To paraphrase Chief Justice Marshall, an "act of the legislature" must be "repugnant to the constitution" in order to be void. Marbury v. Madison, 1 Cranch 137, 177 (1803). When a provision of a statute is unconstitutional, that provision is void, and the Judiciary is therefore not bound by it in a particular case. Here, however, the provisions the majority has excised from the statute are perfectly valid: Congress could pass the identical statute tomorrow and it would be binding on this Court so long as it were administered in compliance with the Sixth Amendment.^[7] Because the statute itself is not repugnant to the Constitution and can by its terms comport with the Sixth Amendment, the Court does not have the constitutional authority to invalidate it.

The precedent on which the Court relies is scant indeed. It can only point to cases in which a provision of law was unconstitutionally extended to or limited to a particular class; in such cases it is necessary either to invalidate the provision or to require the legislature to extend the benefit to an excluded class.^[8] Given the sweeping nature of the remedy ordained today, the majority's assertions that it is proper to engage in an ex ante analysis of congressional intent in order to select in the first instance the statutory provisions to be struck down is contrary to the very purpose of engaging in severability analysis—the Court's remedy expands, rather than limits, judicial power.

There is no justification for extending our severability cases to cover this situation. The SRA and the Guidelines can be read—and are being currently read—in a way that complies with the Sixth Amendment. If Congress wished to amend the statute to enact the majority's vision of how the Guidelines should operate, it would be perfectly free to do so. There is no need to devise a novel and questionable method of invalidating statutory provisions that can be constitutionally applied.

II

Rather than engage in a wholesale rewriting of the SRA, I would simply allow the Government to continue doing what it has done since this Court handed down Blakely—prove any fact that is required to increase a defendant's sentence under the Guidelines to a jury beyond a reasonable doubt. As I have already discussed, a requirement of jury factfinding for certain issues can be implemented without difficulty in the vast majority of cases. See supra, at 276-280.

Indeed, this already appears to be the case. "[T]he Department of Justice already has instituted procedures which would protect the overwhelming majority of future cases from Blakely infirmity. The Department of Justice has issued detailed guidance for every stage of the prosecution from indictment to final sentencing, including alleging facts that would support sentencing enhancements and requiring defendants to waive any potential Blakely rights in plea agreements." Hearings on Blakely 1-2.^[9] Given this experience, I think the Court dramatically overstates the difficulty of implementing this solution.

The majority advances five reasons why the remedy that is already in place will not work. First, the majority points to the statutory text referring to "the court" in arguing that jury factfinding is impermissible. While this text is no doubt evidence that Congress contemplated judicial factfinding, it does not demonstrate that Congress thought that judicial factfinding was so essential that, if forced to choose between a system including jury determinations of certain facts in certain cases on the one hand, and a system in which the Guidelines would cease to restrain the discretion of federal judges on the other, Congress would have selected the latter.

As a textual matter, the word "court" can certainly be read to include a judge's selection of a sentence as supported by a jury verdict—this reading is plausible either as a pure matter of statutory construction or under principles of constitutional avoidance. Ordinarily, "`where a statute is susceptible of two constructions, by one of which grave and doubtful constitutional questions arise and by the other of which such questions are avoided, our duty is to adopt the latter.'" Jones v. United States, 526 U.S. 227, 239 (1999) (quoting United States ex rel. Attorney General v. Delaware & Hudson Co., 213 U.S. 366, 408 (1909)). This principle, which "has for so long been applied by this Court that it is beyond debate," Edward J. DeBartolo Corp. v. Florida Gulf Coast Building & Constr. Trades Council, 485 U. S. 568, 575 (1988), is intended to show respect for Congress by presuming it "legislates in the light of constitutional limitations," Rust v. Sullivan, 500 U. S. 173, 191 (1991).

The Court, however, reverses the ordinary presumption. It interprets the phrase "[t]he court ... shall consider" in 18 U.S.C.A. § 3553(a) (Supp. 2004) to mean: The judge shall consider and impose the appropriate sentence, but the judge shall not be constrained by any findings of a jury. See ante, at 249 (opinion of BREYER, J.) (interpreting the word "court" to mean "`the judge without the jury'"). The Court's narrow reading of the statutory text is unnecessary. Even assuming that the word "court" should be read to mean "judge, and only the judge," a requirement that certain enhancements be supported by jury verdicts leaves the ultimate sentencing decision exclusively within the judge's hands—the judge, and the judge alone, would retain the discretion to sentence the defendant anywhere within the required Guidelines range and within overlapping Guidelines ranges when applicable. See supra, at 278-279. The judge would, no doubt, be limited by the findings of the jury in certain cases, but the fact that such a limitation would be required by the Sixth Amendment in those limited circumstances is not a reason to adopt such a constrained view of an Act of Congress.^[10]

In adopting its constrictive reading of "court," the majority has manufactured a broader constitutional problem than is necessary, and has thereby made necessary the extraordinary remedy it has chosen. I pause, however, to stress that it is not this Court's holding that the Guidelines must be applied consistently with the Sixth Amendment that has made the majority's remedy necessary. Rather, it is the Court's miserly reading of the statutory language that results in "constitutional infirmities." See ante, at 254 (opinion of BREYER, J.)

Second, the Court argues that simply applying Blakely to the Guidelines would make "real conduct" sentencing more difficult. While that is perhaps true in some cases, judges could always consider relevant conduct obtained from a presentence report pursuant to 18 U.S.C. § 3661 and USSG § 6A1.1 in selecting a sentence within a Guidelines range, and of course would be free to consider any such circumstances in cases in which the defendant pleads guilty and waives his Blakely rights. Further, in many cases the Government could simply prove additional facts to a jury beyond a reasonable doubt—as it has been doing in some cases since Apprendi—or the court could use bifurcated proceedings in which the relevant conduct is proved to a jury after it has convicted the defendant of the underlying crime.

The majority is correct, however, that my preferred holding would undoubtedly affect "real conduct" sentencing in certain cases. This is so because the goal of such sentencing—increasing a defendant's sentence on the basis of conduct not proved at trial—is contrary to the very core of Apprendi. That certain applications of "relevant conduct" sentencing are unconstitutional should not come as a complete surprise to Congress: The House Report recognized that "real offense" sentencing could pose constitutional difficulties. H. R. Rep. No. 98-1017, p. 98 (1984). In reality, the majority's concerns about relevant conduct are nothing more than an objection to Apprendi itself, an objection that this Court rejected in Parts I-III, ante (opinion of STEVENS, J.).

Further, the Court does not explain how its proposed remedy will ensure that judges take real conduct into account. While judges certainly may do so in their discretion under § 3553(a), there is no indication as to how much or to what extent "relevant conduct" should matter under the majority's regime. Nor is there any meaningful standard by which appellate courts may review a sentencing judge's "relevant conduct" determination—only a general "reasonableness" inquiry that may discourage sentencing judges from considering such conduct altogether. The Court's holding thus may do just as much damage to real conduct sentencing as would simply requiring the Government to follow the Guidelines consistent with the Sixth Amendment.

Third, the majority argues that my remedy would make sentencing proceedings far too complex. But of the very small number of cases in which a Guidelines sentence would implicate the Sixth Amendment, see supra, at 275-276, most involve drug quantity determinations, firearm enhancements, and other factual findings that can readily be made by juries. I am not blind to the fact that some cases, such as fraud prosecutions, would pose new problems for prosecutors and trial judges. See ante, at 252-253 (opinion of BREYER, J.). In such cases, I am confident that federal trial judges, assisted by capable prosecutors and defense attorneys, could have devised appropriate procedures to impose the sentences the Guidelines envision in a manner that is consistent with the Sixth Amendment. We have always trusted juries to sort through complex facts in various areas of law. This may not be the most efficient system imaginable, but the Constitution does not permit efficiency to be our primary concern. See Blakely v. Washington, 542 U. S., at 312-313.

Fourth, the majority assails my reliance on plea bargaining. The Court claims that I cannot discount the effect that applying Blakely to the Guidelines would have on plea-bargained cases, since the specter of Blakely will affect those cases. However, the majority's decision suffers from the same problem to a much greater degree. Prior to the Court's decision to strike the mandatory feature of the Guidelines, prosecutors and defendants alike could bargain from a position of reasonable confidence with respect to the sentencing range into which a defendant would likely fall. The majority, however, has eliminated the certainty of expectations in the plea process. And, unlike my proposed remedy, which would potentially affect only a fraction of plea bargains, the uncertainty resulting from the Court's regime change will infect the entire universe of guilty pleas which occur in 97% of all federal prosecutions.

The majority also argues that applying Blakely to the Guidelines would allow prosecutors to exercise "a power the Sentencing Act vested in judges," ante, at 257 (opinion of BREYER, J.), by giving prosecutors the choice whether to "charge" a particular fact. Under the remedy I favor, however, judges would still be able to reject factually false plea agreements under USSG § 6B1.2(a), and could still consider relevant information about the offense and the offender in every single case. Judges could consider such characteristics as an aid in selecting the appropriate sentence within the Guidelines range authorized by the jury verdict, determining the defendant's criminal history level, reducing a defendant's sentence, or justifying discretionary departures from the applicable Guidelines range. The Court is therefore incorrect when it suggests that requiring a supporting jury verdict for certain enhancements in certain cases would place certain sentencing factors "beyond the reach of the judge entirely." See ante, at 257 (opinion of BREYER, J.).

Moreover, the premise on which the Court's argument is based—that the Guidelines as currently written prevent fact bargaining and therefore diminish prosecutorial power—is probably not correct. As one commentator has noted,

"prosecutors exercise nearly as much control when guidelines tie sentences to so-called `real-offense' factors.... One might reasonably assume those factors are outside of prosecutors' control, but experience with the Federal Sentencing Guidelines suggests otherwise; when necessary, the litigants simply bargain about what facts will (and won't) form the basis for sentencing. It seems to be an iron rule: guidelines sentencing empowers prosecutors, even where the guidelines' authors try to fight that tendency." Stuntz, Plea Bargaining and Criminal Law's Disappearing Shadow, 117 Harv. L. Rev. 2548, 2559-2560 (2004) (footnote omitted).

Not only is fact bargaining quite common under the current system, it is also clear that prosecutors have substantial bargaining power.^[11] And surely, contrary to the Court's response to this dissent, ante, at 256-257 (opinion of BREYER, J.), a prosecutor who need only prove an enhancing fact by a preponderance of the evidence has more bargaining power than if required to prove the same fact beyond a reasonable doubt.

Finally, the majority argues that my solution would require a different burden of proof for enhancements above the maximum authorized by the jury verdict and for reductions. This is true because the requirement that guilt be established by proof beyond a reasonable doubt is a constitutional mandate. However, given the relatively few reductions available in the Guidelines and the availability of judicial discretion within the applicable range, this is unlikely to have more than a minimal effect.

In sum, I find unpersuasive the Court's objections to allowing Congress to decide in the first instance whether the Guidelines should be converted from a mandatory into a discretionary system. Far more important than those objections is the overwhelming evidence that Congress has already considered, and unequivocally rejected, the regime that the Court endorses today.

III

Even under the Court's innovative approach to severability analysis when confronted with unconstitutional applications of a statute, its opinion is unpersuasive. It assumes that this Court's only inquiry is to "decide whether we would deviate less radically from Congress' intended system (1) by superimposing the constitutional requirement announced today or (2) through elimination of some provisions of the statute." Ante, at 247 (opinion of BREYER, J.). I will assume, consistently with the majority, that in this exercise we should never use our "remedial powers to circumvent the intent of the legislature," Califano v. Westcott, 443 U. S. 76, 94 (1979) (Powell, J., concurring in part and dissenting in part), and that we must not create "a program quite different from the one the legislature actually adopted," Sloan v. Lemon, 413 U. S. 825, 834 (1973).

In the context of this framework, in order to justify "excising" 18 U. S. C. A. §§ 3553(b)(1) (Supp. 2004) and 3742(e) (main ed. and Supp. 2004), the Court has the burden of showing that Congress would have preferred the remaining system of discretionary Sentencing Guidelines to not just the remedy I would favor, but also to any available alternative, including the alternative of total invalidation, which would give Congress a clean slate on which to write an entirely new law. The Court cannot meet this burden because Congress has already considered and overwhelmingly rejected the system it enacts today. In doing so, Congress revealed both an unmistakable preference for the certainty of a binding regime and a deep suspicion of judges' ability to reduce disparities in federal sentencing. A brief examination of the SRA's history reveals the gross impropriety of the remedy the Court has selected.

History of Sentence Reform Efforts:

In the mid-1970's, Congress began to study the numerous problems attendant to indeterminate sentencing in the federal criminal justice system. After nearly a decade of review, Congress in 1984 decided that the system needed a comprehensive overhaul. The elimination of sentencing disparity, which Congress determined was chiefly the result of a discretionary sentencing regime, was unquestionably Congress' principal aim. See Feinberg, Federal Criminal Sentencing Reform: Congress and the United States Sentencing Commission, 28 Wake Forest L. Rev. 291, 295-296 (1993) ("The first and foremost goal of the sentencing reform effort was to alleviate the perceived problem of federal criminal sentencing disparity.... Quite frankly, all other considerations were secondary"); see also Breyer, Federal Sentencing Guidelines Revisited, 2 Fed. Sentencing Rptr. 180 (1999) ("In seeking `greater fairness,' Congress, acting in bipartisan fashion, intended to respond to complaints of unreasonable disparity in sentencing—that is, complaints that differences among sentences reflected not simply different offense conduct or different offender history, but the fact that different judges imposed the sentences" (emphases added)). As Senator Hatch, a central participant in the reform effort, has explained: "The discretion that Congress had conferred for so long upon the judiciary and the parole authorities was at the heart of sentencing disparity." The Role of Congress in Sentencing: The United States Sentencing Commission, Mandatory Minimum Sentences, and the Search for a Certain and Effective Sentencing System, 28 Wake Forest L. Rev. 185, 187 (1993) (hereinafter Hatch) (emphasis added).

Consequently, Congress explicitly rejected as a model for reform the various proposals for advisory guidelines that had been introduced in past Congresses. One example of such legislation was the bill introduced in 1977 by Senators Kennedy and McClellan, S. 1437, 95th Cong., 1st Sess. (as reported by the Senate Judiciary Committee on Nov. 15, 1977) (hereinafter S. 1437), which allowed judges to impose sentences based on the characteristics of the individual defendant and granted judges substantial discretion to depart from recommended guidelines sentences. See Stith & Koh, The Politics of Sentencing Reform: The Legislative History of the Federal Sentencing Guidelines, 28 Wake Forest L. Rev. 223, 238 (1993) (hereinafter Stith & Koh). That bill never became law and was refined several times between 1977 and 1984: Each of those refinements made the regime more, not less, restrictive on trial judges' discretion in sentencing.^[12]

Passage of the Sentencing Reform Act of 1984:

Congress' preference for binding guidelines was evident in the debate over passage of the SRA itself, which was predicated entirely on the move from a discretionary guidelines system to the mandatory system the Court strikes down today. The SRA was the product of competing versions of sentencing reform legislation: the House bill, H. R. 6012, 98th Cong., 2d Sess., authorized the creation of discretionary guidelines whereas the Senate bill, S. 668, 98th Cong., 2d Sess., provided for binding guidelines and de novo appellate review. The House was splintered regarding whether to make the Guidelines binding on judges, but the vote in the Senate was an overwhelming 85 to 3 in favor of binding Guidelines. 130 Cong. Rec. 1649 (1984); see generally Stith & Koh 261-266. Eventually, the House substituted the Senate version for H. R. 6012, and the current system of mandatory Guidelines became law. 130 Cong. Rec. 29730 (1984).

The text of the law that actually passed Congress (including §§ 3553(b)(1) and 3742(e)) should be more than sufficient to demonstrate Congress' unmistakable commitment to a binding Guidelines system. That text requires the sentencing judge to impose the sentence dictated by the Guidelines ("[T]he court shall impose a sentence of the kind, and within the range" provided in the Guidelines unless there is a circumstance "not adequately taken into consideration by the" Guidelines), and § 3742(e) gives § 3553(b)(1) teeth by instructing judges that any sentence outside of the Guidelines range without adequate explanation will be overturned on appeal.^[13] Congress' chosen regime was carefully designed to produce uniform compliance with the Guidelines. Congress surely would not have taken the pains to create such a regime had it found the Court's system of discretionary guidelines acceptable in any way.

The accompanying Senate Report and floor debate make plain what should be obvious from the structure of the statute: Congress refused to accept the discretionary system that the Court implausibly deems most consistent with congressional intent.^[14] In other words, given the choice between the statute created by the Court today or a clean slate on which to write a wholly different law, Congress undoubtedly would have selected the latter.

Congress' Method of Reducing Disparities:

The notion that Congress had any confidence that judges would reduce sentencing disparities by considering relevant conduct — an idea that is championed by the Court, ante, at 253-254 (opinion of BREYER, J.) — either ignores or misreads the political environment in which the SRA passed. It is true that the SRA instructs sentencing judges to consider real offense and offender characteristics, 28 U.S.C.A. § 994 (main ed. and Supp. 2004), but Congress only wanted judges to consider those characteristics within the limits of a mandatory system.^[15] The Senate Report on which the Court relies, see ante, at 249-250, clearly concluded that the existence of sentencing disparities "can be traced directly to the unfettered discretion the law confers on those judges . . . responsible for imposing and implementing the sentence." S. Rep. No. 98-225, p. 38 (1983). Even in a system in which judges could not impose sentences based on "relevant conduct" determinations (absent a plea agreement or supporting jury findings), sentences would still be every bit as certain and uniform as in the status quo — at most, the process for imposing those sentences would be more complex. The same can hardly be said of the Court's chosen system, in which all federal sentencing judges, in all cases, regain the unconstrained discretion Congress eliminated in 1984.

The Court's conclusion that Congress envisioned a sentencing judge as the centerpiece of its effort to reduce disparities is remarkable given the context of the broader legislative debate about what entity would be responsible for drafting the Guidelines under the SRA. The House version of the bill preferred the Guidelines to be written by the Judicial Conference of the United States — the House Report accompanying that bill argued that judges had vast experience in sentencing and would best be able to craft a system capable of providing sentences based on real conduct without excessive disparity. See H. R. Rep. No. 98-1017, at 93-94. Those in the Senate majority, however, favored an independent Commission. They did so, whether rightly or wrongly, based on a belief that federal judges could not be trusted to impose fair and uniform sentences. See, e. g., 130 Cong. Rec. 976 (1984) (remarks of Sen. Laxalt) ("The present problem with disparity in sentencing . . . stems precisely from the failure of [f]ederal judges — individually and collectively — to sentence similarly situated defendants in a consistent, reasonable manner. There is little reason to believe that judges will now begin to do what they have failed to do in the past"). And, at the end of the debate, the few remaining Members in the minority recognized that the battle to empower judges with more discretion had been lost. See, e. g., id., at 973 (remarks of Sen. Mathias) (arguing that "[t]he proponents of the bill ... argue in essence that judges cannot be trusted. You cannot trust a judge ... you must not trust a judge"). I find it impossible to believe that a Congress in which these sentiments prevailed would have ever approved of the discretionary sentencing regime the Court enacts today.

Congressional Activity Since 1984:

Congress has not wavered in its commitment to a binding system of Sentencing Guidelines. In fact, Congress has rejected each and every attempt to loosen the rigidity of the Guidelines or vest judges with more sentencing options. See Hatch 189 ("In ensuing years, Congress would maintain its adherence to the concept of binding guidelines by consistently rejecting efforts to make the guidelines more discretionary"). Most recently, Congress' passage of the Prosecutorial Remedies and Other Tools to end the Exploitation of Children Today Act of 2003 (PROTECT Act), Pub. L. 108-21, 117 Stat. 650, reinforced the mandatory nature of the Guidelines by expanding de novo review of sentences to include all departures from the Guidelines and by directing the Commission to limit the number of available departures. The majority admits that its holding has made the PROTECT Act irrelevant. See ante, at 261 (opinion of BREYER, J.) (admitting that after the Court's remedy, the PROTECT Act's provisions "have ceased to be relevant"). Even a cursory reading of the legislative history of the PROTECT Act reveals the absurdity of the claim that Congress would find acceptable, under any circumstances, the Court's restoration of judicial discretion through the facial invalidation of §§ 3553(b)(1) and 3742(e).^[16] In sum, despite Congress' unequivocal demand that the Guidelines operate as a binding system, and in the name of avoiding any reduction in the power of the sentencing judge vis-à-vis the jury (a subject to which Congress did not speak), the majority has erased the heart of the SRA and ignored in their entirety all of the Legislative Branch's postenactment expressions of how the Guidelines are supposed to operate.

The majority's answer to this overwhelming history is that retaining a mandatory Guidelines system "is not a choice that remains open" given our holding that Blakely applies to the Guidelines. Ante, at 265. This argument—essentially, that the Apprendi rule makes determinate sentencing unconstitutional — has been advanced repeatedly since Apprendi. See, e. g., 530 U. S., at 549-554 (O'CONNOR, J., dissenting); Blakely, 542 U. S., at 314 (O'CONNOR, J., dissenting); id., at 345-346 (BREYER, J., dissenting). These prophecies were self fulfilling. It is not Apprendi that has brought an end to determinate sentencing. This Court clearly had the power to adopt a remedy that both complied with the Sixth Amendment and also preserved a determinate sentencing regime in which judges make regular factual determinations regarding a defendant's sentence. It has chosen instead to exaggerate the constitutional problem and to expand the scope of judicial invalidation far beyond that which is even arguably necessary. Our holding that Blakely applies to the Sentencing Guidelines did not dictate the Court's unprecedented remedy.

IV

As a matter of policy, the differences between the regime enacted by Congress and the system the Court has chosen are stark. Were there any doubts about whether Congress would have preferred the majority's solution, these are sufficient to dispel them. First, Congress' stated goal of uniformity is eliminated by the majority's remedy. True, judges must still consider the sentencing range contained in the Guidelines, but that range is now nothing more than a suggestion that may or may not be persuasive to a judge when weighed against the numerous other considerations listed in 18 U. S. C. A. § 3553(a) (main ed. and Supp. 2004). The result is certain to be a return to the same type of sentencing disparities Congress sought to eliminate in 1984. Prior to the PROTECT Act, rates of departure from the applicable Guidelines sentence (via upward or downward departure) varied considerably depending upon the Circuit in which one was sentenced. See Sourcebook 53-55 (Table 26) (showing that 76.6% of sentences in the Fourth Circuit were within the applicable Guidelines range, whereas only 48.8% of sentences in the Ninth Circuit fell within the range). Those disparities will undoubtedly increase in a discretionary system in which the Guidelines are but one factor a judge must consider in sentencing a defendant within a broad statutory range.

Moreover, the Court has neglected to provide a critical procedural protection that existed prior to the enactment of a binding Guidelines system. Before the SRA, the sentencing judge had the discretion to impose a sentence that designated a minimum term "at the expiration of which the prisoner shall become eligible for parole." 18 U. S. C. § 4205(b) (1982 ed.) (repealed by Pub. L. 98-473, § 218(a)(5), 98 Stat. 2027). Sentencing judges had the discretion to reduce a minimum term of imprisonment upon the recommendation of the Bureau of Prisons. § 4205(g). Through these provisions and others, see generally §§ 4201-4215, all of which were effectively repealed in 1984, it was the Parole Commission — not the sentencing judge — who was ultimately responsible for determining the length of each defendant's real sentence. See, e. g., S. Rep. No. 98-225, at 38. Prior to the Guidelines regime, the Parole Commission was designed to reduce sentencing disparities and to provide a check for defendants who had received excessive sentences. Today, the Court reenacts the discretionary Guidelines system that once existed without providing this crucial safety net.

Other concerns are likely to arise. Congress' demand in the PROTECT Act that departures from the Guidelines be closely regulated and monitored is eviscerated — for there can be no "departure" from a mere suggestion. How will a judge go about determining how much deference to give to the applicable Guidelines range? How will a court of appeals review for reasonableness a district court's decision that the need for "just punishment" and "adequate deterrence to criminal conduct" simply outweighs the considerations contemplated by the Sentencing Commission? See 18 U.S.C. §§ 3553(a)(2)(A)-(B). What if a sentencing judge determines that a defendant's need for "educational or vocational training, medical care, or other correctional treatment in the most effective manner," § 3553(a)(2)(D), requires disregarding the stiff Guidelines range Congress presumably preferred? These questions will arise in every case in the federal system under the Court's system. Regrettably, these are exactly the sort of questions Congress hoped that sentencing judges would not ask after the SRA.

The consequences of such a drastic change — unaided by the usual processes of legislative deliberation — are likely to be sweeping. For example, the majority's unnecessarily broad remedy sends every federal sentence back to the drawing board, or at least into the novel review for "reasonableness," regardless of whether those individuals' constitutional rights were violated. It is highly unlikely that the mere application of "prudential doctrines" will mitigate the consequences of such a gratuitous change.

The majority's remedy was not the inevitable result of the Court's holding that Blakely applies to the Guidelines. Neither Apprendi, nor Blakely, nor these cases made determinate sentencing unconstitutional.^[17] Merely requiring all applications of the Guidelines to comply with the Sixth Amendment would have allowed judges to distinguish harmless error from error requiring correction, would have required no more complicated procedures than the procedural regime the majority enacts today, and, ultimately, would have left most sentences intact.

Unlike a rule that would merely require judges and prosecutors to comply with the Sixth Amendment, the Court's systematic overhaul turns the entire system on its head in every case, and, in so doing, runs contrary to the central purpose that motivated Congress to act in the first instance. Moreover, by repealing the right to a determinate sentence that Congress established in the SRA, the Court has effectively eliminated the very constitutional right Apprendi sought to vindicate. No judicial remedy is proper if it is "not commensurate with the constitutional violation to be repaired." Hills v. Gautreaux, 425 U. S. 284, 294 (1976). The Court's system fails that test, frustrates Congress' principal goal in enacting the SRA, and violates the tradition of judicial restraint that has heretofore limited our power to overturn validly enacted statutes.

I respectfully dissent.

JUSTICE SCALIA, dissenting in part.

I join the portions of the opinion of the Court that are delivered by JUSTICE STEVENS. I also join JUSTICE STEVENS's dissent, with the exception of Part III^[1] and footnote 17. I write separately mainly to add some comments regarding the change that the remedial majority's handiwork has wrought (or perhaps — who can tell? — has not wrought) upon appellate review of federal sentencing.

The remedial majority takes as the North Star of its analysis the fact that Congress enacted a "judge-based sentencing system." Ante, at 265 (opinion of BREYER, J.). That seems to me quite misguided. Congress did indeed expect judges to make the factual determinations to which the Guidelines apply, just as it expected the Guidelines to be mandatory. But which of those expectations was central to the congressional purpose is not hard to determine. No headline describing the Sentencing Reform Act of 1984 (Act) would have read "Congress reaffirms judge-based sentencing" rather than "Congress prescribes standardized sentences." JUSTICE BREYER's opinion for the Court repeatedly acknowledges that the primary objective of the Act was to reduce sentencing disparity.^[2] Inexplicably, however, the opinion concludes that the manner of achieving uniform sentences was more important to Congress than actually achieving uniformity — that Congress was so attached to having judges determine "real conduct" on the basis of bureaucratically prepared, hearsay-riddled presentence reports that it would rather lose the binding nature of the Guidelines than adhere to the old-fashioned process of having juries find the facts that expose a defendant to increased prison time. See ante, at 253-254, 265. The majority's remedial choice is thus wonderfully ironic: In order to rescue from nullification a statutory scheme designed to eliminate discretionary sentencing, it discards the provisions that eliminate discretionary sentencing.

That is the plain effect of the remedial majority's decision to excise 18 U. S. C. A. § 3553(b)(1) (Supp. 2004). See ante, at 259. District judges will no longer be told they "shall impose a sentence ... within the range" established by the Guidelines. § 3553(b)(1). Instead, under § 3553(a), they will need only to "consider" that range as one of many factors, including "the need for the sentence ... to provide just punishment for the offense," § 3553(a)(2)(A) (main ed.), "to afford adequate deterrence to criminal conduct," § 3553(a)(2)(B), and "to protect the public from the further crimes of the defendant," § 3553(a)(2)(C). The statute provides no order of priority among all those factors, but since the three just mentioned are the fundamental criteria governing penology, the statute — absent the mandate of § 3553(b)(1) — authorizes the judge to apply his own perceptions of just punishment, deterrence, and protection of the public even when these differ from the perceptions of the Commission members who drew up the Guidelines. Since the Guidelines are not binding, in order to comply with the (oddly) surviving requirement that the court set forth "the specific reason for the imposition of a sentence different from that described" in the Guidelines, § 3553(c)(2), the sentencing judge need only state that "this court does not believe that the punishment set forth in the Guidelines is appropriate for this sort of offense."^[3] That is to say, district courts have discretion to sentence anywhere within the ranges authorized by statute — much as they were generally able to do before the Guidelines came into being. To be sure, factor (6) is "the need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct," § 3553(a)(2)(6) (main ed.), but this would require a judge to adhere to the Guidelines only if all other judges had to adhere to the Guidelines (which they certainly do not, as the Court holds today) or if all other judges could at least be expected to adhere to the Guidelines (which they certainly cannot, given the notorious unpopularity of the Guidelines with many district judges). Thus, logic compels the conclusion that the sentencing judge, after considering the recited factors (including the Guidelines), has full discretion, as full as what he possessed before the Act was passed, to sentence anywhere within the statutory range. If the majority thought otherwise — if it thought the Guidelines not only had to be "considered" (as the amputated statute requires) but had generally to be followed — its opinion would surely say so.^[4]

As frustrating as this conclusion is to the Act's purpose of uniform sentencing, it at least establishes a clear and comprehensible regime — essentially the regime that existed before the Act became effective. That clarity is eliminated, however, by the remedial majority's surgery on 18 U. S. C. A. §3742 (main ed. and Supp. 2004), the provision governing appellate review of sentences. Even the most casual reading of this section discloses that its purpose — its only purpose — is to enable courts of appeals to enforce conformity with the Guidelines. All of the provisions of that section that impose a review obligation beyond what existed under prior law^[5] are related to the district judge's obligations under the Guidelines. If the Guidelines are no longer binding, one would think that the provision designed to ensure compliance with them would, in its totality, be inoperative. The Court holds otherwise. Like a black-robed Alexander cutting the Gordian knot, it simply severs the purpose of the review provisions from their text, holding that only subsection (e), which sets forth the determinations that the court of appeals must make, is inoperative, whereas all the rest of § 3742 subsists — including, mirabile dictu, subsection (f), entitled "Decision and disposition," which tracks the determinations required by the severed subsection (e) and specifies what disposition each of those determinations is to produce. This is rather like deleting the ingredients portion of a recipe and telling the cook to proceed with the preparation portion.^[6]

Until today, appellate review of sentencing discretion has been limited to instances prescribed by statute. Before the Guidelines, federal appellate courts had little experience reviewing sentences for anything but legal error. "[W]ell-established doctrine," this Court said, "bars [appellate] review of the exercise of sentencing discretion." Dorszynski v. United States, 418 U. S. 424, 443 (1974). "[O]nce it is determined that a sentence is within the limitations set forth in the statute under which it is imposed, appellate review is at an end." Id., at 431-432 (citing cases). When it established the Guidelines regime, Congress expressly provided for appellate review of sentences in specified circumstances, but the Court has been appropriately chary of aggrandizement, refusing to treat § 3742 as a blank check to appellate courts. Thus, in 1992, the Court recognized that Congress's grant of "limited appellate review of sentencing decisions ... did not alter a court of appeals' traditional deference to a district court's exercise of its sentencing discretion." Williams v. United States, 503 U. S. 193, 205 (emphasis added). Notwithstanding § 3742, much remained off limits to the courts of appeals: "The selection of the appropriate sentence from within the guideline range, as well as the decision to depart from the range in certain circumstances, are decisions that are left solely to the sentencing court." Ibid. (emphasis added). Similarly, in 1996, the Court took pains to note that the § 3742 power to engage in "limited appellate review" of Guidelines departures did not "vest in appellate courts wide-ranging authority over district court sentencing decisions." Koon v. United States, 518 U. S. 81, 97. The Court repeated its caution that "`[t]he development of the guideline sentencing regime'" did not allow appellate review "`except to the extent specifically directed by statute.'" Ibid. (quoting Williams, supra, at 205).

Today's remedial opinion does not even pretend to honor this principle that sentencing discretion is unreviewable except pursuant to specific statutory direction. The discussion of appellate review begins with the declaration that, "despite the absence of § 3553(b)(1) (Supp. 2004), the Act continues to provide for appeals from sentencing decisions (irrespective of whether the trial judge sentences within or outside the Guidelines range ...)," ante, at 260 (citing §§ 3742(a) and (b)); and the opinion later announces that the standard of review for all such appeals is "unreasonableness," ante, at 261, 264-265. This conflates different and distinct statutory authorizations of appeal and elides crucial differences in the statutory scope of review. Section 3742 specifies four different kinds of appeal,^[7] setting forth for each the grounds of appeal permitted to the defendant and the Government (§§ 3742(a) and (b)), the manner in which each ground should be considered (§ 3742(e)), and the permissible dispositions (§ 3742(f)). There is no one-size-fits-all "unreasonableness" review. The power to review a sentence for reasonableness arises only when the sentencing court has departed from "the applicable guideline range." § 3742(f)(2); cf. United States v. Soltero-Lopez, 11 F. 3d 18, 19 (CA1 1993) (Breyer, C. J.) ("[T]he sentencing statutes ... provide [a defendant] with only a very narrow right of appeal" because the power "to set aside a departure that is `unreasonable'" appears "in the context of other provisions that permit defendants to appeal only upward . . . departures"). This Court has expressly rejected the proposition that there may be a "reasonable[ness]" inquiry when a sentence is imposed as a result of an incorrect application of the Guidelines. See Williams, supra, at 201.

The Court claims that "a statute that does not explicitly set forth a standard of review may nonetheless do so implicitly." Ante, at 260 (opinion of BREYER, J.). Perhaps so. But we have before us a statute that does explicitly set forth a standard of review. The question is, when the Court has severed that standard of review (contained in § 3742(e)), does it make any sense to look for some congressional "implication" of a different standard of review in the remnants of the statute that the Court has left standing? Only in Wonderland. (This may explain in part why, as JUSTICE STEVENS's dissent correctly observes, ante, at 282, none of the numerous persons and organizations filing briefs as parties or amici in these cases — all of whom filed this side of the looking-glass — proposed, or I think even imagined, the remedial majority's wonderful disposition.) Unsurprisingly, none of the three cases cited by the Court used the power of implication to fill a gap created by the Court's own removal of an explicit standard.^[8] The Court's need to create a new, "implied" standard of review—however "linguistically" "fair," ante, at 262—amounts to a confession that it has exceeded its powers. According to the "well established" standard for severability, the unconstitutional part of a statute "may be dropped if what is left is fully operative as a law." Alaska Airlines, Inc. v. Brock, 480 U.S. 678, 684 (1987) (emphasis added and internal quotation marks omitted). Severance is not possible "if the balance of the legislation is incapable of functioning independently." Ibid. The Court's need to supplement the text that remains after severance suggests that it is engaged in "redraft[ing] the statute" rather than just implementing the valid portions of it. United States v. Treasury Employees, 513 U.S. 454, 479, and n. 26 (1995); see also id., at 502, and n. 8 (REHNQUIST, C. J., dissenting); Reno v. American Civil Liberties Union, 521 U.S. 844, 884-885 (1997).

Even assuming that the Court ought to be inferring standards of review to stanch the bleeding created by its aggressive severance of § 3742(e), its "unreasonableness" standard is not, as it claims, consistent with the "related statutory language" or with "appellate sentencing practice during the last two decades." Ante, at 260, 262. As already noted, sentences within the Guidelines range have not previously been reviewed for reasonableness. Indeed, the very concept of having a unitary standard of review for all kinds of appeals authorized by §§ 3742(a) and (b) finds no support in statutory language or established practice of the last two decades. Although a "reasonableness" standard did appear in § 3742(e)(3) until 2003, it never extended beyond review of deliberate departures from the Guidelines range. See 18 U.S.C. § 3742(e)(3) (2000 ed.); see also §§ 3742(f)(2)(A), (B) (prescribing how to dispose on appeal of a sentence that is "outside the applicable guideline range and is unreasonable"). According to the statistics cited by the Court, that standard applied to only 16.7% of federal sentencing appeals in 2002, see ante, at 262 (opinion of BREYER, J.), but the Court would now have it apply across the board to all sentencing appeals, even to sentences within "the applicable guideline range," where there is no legal error or misapplication of the Guidelines.

There can be no doubt that the Court's severability analysis has produced a scheme dramatically different from anything Congress has enacted since 1984. Sentencing courts are told to "provide just punishment" (among other things), and appellate courts are told to ensure that district judges are not "unreasonable." The worst feature of the scheme is that no one knows—and perhaps no one is meant to know—how advisory Guidelines and "unreasonableness" review will function in practice. The Court's description of what it anticipates is positively Delphic: "These features of the remaining system . . . continue to move sentencing in Congress' preferred direction, helping to avoid excessive sentencing disparities while maintaining flexibility sufficient to individualize sentences where necessary. We can find no feature of the remaining system that tends to hinder, rather than to further, these basic objectives." Ante, at 264-265 (citation omitted).

As I have suggested earlier, any system which held it per se unreasonable (and hence reversible) for a sentencing judge to reject the Guidelines is indistinguishable from the mandatory Guidelines system that the Court today holds unconstitutional. But the remedial majority's gross exaggerations (it says that the "practical standard of review" it prescribes is "already familiar to appellate courts" and "consistent with appellate sentencing practice during the last two decades," ante, at 261, 262)^[9] may lead some courts of appeals to conclude—may indeed be designed to lead courts of appeals to conclude—that little has changed. Bear in mind that one of the most significant features of the remedial majority's scheme of "unreasonableness" review is that it requires courts of appeals to evaluate each sentence individually for reasonableness, rather than apply the cookie-cutter standards of the mandatory Guidelines (within the correct Guidelines range, affirm; outside the range without adequate explanation, vacate and remand). A court of appeals faced with this daunting prospect might seek refuge in the familiar and continue (as the remedial majority invites, though the merits majority forbids) the "appellate sentencing practice during the last two decades," ante, at 262 (opinion of BREYER, J.). At the other extreme, a court of appeals might handle the new workload by approving virtually any sentence within the statutory range that the sentencing court imposes, so long as the district judge goes through the appropriate formalities, such as expressing his consideration of and disagreement with the Guidelines sentence. What I anticipate will happen is that "unreasonableness" review will produce a discordant symphony of different standards, varying from court to court and judge to judge, giving the lie to the remedial majority's sanguine claim that "no feature" of its avant-garde Guidelines system will "ten[d] to hinder" the avoidance of "excessive sentencing disparities." Ante, at 265.

In Blakely v. Washington, 542 U.S. 296 (2004), the four dissenting Justices accused the Court of ignoring "the havoc it is about to wreak on trial courts across the country." Id., at 324 (opinion of O'CONNOR, J.). And that harsh assessment, of course, referred to just a temporary and unavoidable uncertainty, until the Court could get before it a case properly presenting the constitutionality of the mandatory Guidelines. Today, the same Justices wreak havoc on federal district and appellate courts quite needlessly, and for the indefinite future. Will appellate review for "unreasonableness" preserve de facto mandatory Guidelines by discouraging district courts from sentencing outside Guidelines ranges? Will it simply add another layer of unfettered judicial discretion to the sentencing process? Or will it be a mere formality, used by busy appellate judges only to ensure that busy district judges say all the right things when they explain how they have exercised their newly restored discretion? Time may tell, but today's remedial majority will not.

I respectfully dissent.

JUSTICE THOMAS, dissenting in part.

I join JUSTICE STEVENS' opinion for the Court, but I dissent from JUSTICE BREYER's opinion for the Court. While I agree with JUSTICE STEVENS' proposed remedy and much of his analysis, I disagree with his restatement of severability principles and reliance on legislative history, and thus write separately.

The Constitution prohibits allowing a judge alone to make a finding that raises the sentence beyond the sentence that could have lawfully been imposed by reference to facts found by the jury or admitted by the defendant. Application of the Federal Sentencing Guidelines resulted in impermissible factfinding in Booker's case, but not in Fanfan's. Thus Booker's sentence is unconstitutional, but Fanfan's is not. Rather than applying the usual presumption in favor of severability, and leaving the Guidelines standing insofar as they may be applied without any constitutional problem, the remedial majority converts the Guidelines from a mandatory system to a discretionary one. The majority's solution fails to tailor the remedy to the wrong, as this Court's precedents require.

I

When a litigant claims that a statute is unconstitutional as applied to him, and the statute is in fact unconstitutional as applied, we normally invalidate the statute only as applied to the litigant in question. We do not strike down the statute on its face. In the typical case, "we neither want nor need to provide relief to nonparties when a narrower remedy will fully protect the litigants." United States v. Treasury Employees, 513 U.S. 454, 478 (1995); see also Renne v. Geary, 501 U.S. 312, 323-324 (1991); Board of Trustees of State Univ. of N.Y. v. Fox, 492 U.S. 469, 484-485 (1989); Brockett v. Spokane Arcades, Inc., 472 U.S. 491, 501-504 (1985). Absent an exception such as First Amendment overbreadth, we will facially invalidate a statute only if the plaintiff establishes that the statute is invalid in all of its applications. United States v. Salerno, 481 U.S. 739, 745 (1987).

Booker's case presents an as-applied challenge. Booker challenges Guidelines enhancements that, based on factfinding by a judge alone, raised his sentence above the range legally mandated for his base offense level, determined by reference to the jury verdict. In effect, he contends that the Guidelines supporting the enhancements, and the Sentencing Reform Act of 1984 (SRA) that makes the Guidelines enhancements mandatory, were unconstitutionally applied to him. (Fanfan makes no similar contention, as he seeks to uphold the District Court's application of the Guidelines.)

A provision of the SRA, 18 U.S.C.A. § 3553(b)(1) (Supp. 2004), commands that the court "shall impose a sentence of the kind, and within the range, referred to in subsection (a)(4)," which in turn refers to the Guidelines. (Emphasis added.) The Court reasons that invalidating § 3553(b)(1) would render the Guidelines nonbinding and therefore constitutional. Hence, it concludes, § 3553(b)(1) must fall on its face.^[1]

The majority's excision of § 3553(b)(1) is at once too narrow and too broad. It is too narrow in that it focuses only on § 3553(b)(1), when Booker's unconstitutional sentence enhancements stemmed not from § 3553(b)(1) alone, but from the combination of § 3553(b)(1) and individual Guidelines. Specifically, in Booker's case, the District Court increased the base offense level^[2] under these Guidelines^[3]: USSG § 1B1.3(a)(2), which instructs that the base offense level shall (for certain offenses) take into account all acts "that were part of the same course of conduct or common scheme or plan as the offense of conviction"; § 2D1.1(c)(2), which sets the offense level for 500g to 1.5kg of cocaine base at 36; and § 3C1.1, which provides for a two-level increase in the offense level for obstruction of justice. The court also implicitly applied § 1B1.1, which provides general instructions for applying the Guidelines, including determining the base offense level and applying appropriate adjustments; § 1B1.11(b)(2), which requires that "[t]he Guidelines Manual in effect on a particular date shall be applied in its entirety"; § 6A1.3(b) p. s.,^[4] which provides that "[t]he court shall resolve disputed sentencing factors at a sentencing hearing in accordance with Rule 32(c)(1), Fed. R. Crim. P."; and Rule 32(c)(1),^[5] which in turn provided:

"At the sentencing hearing, the court ... must rule on any unresolved objections to the presentence report.... For each matter controverted, the court must make either a finding on the allegation or a determination that no finding is necessary because the controverted matter will not be taken into account in, or will not affect, sentencing."

Section 3553(b)(1), the listed Guidelines and policy statement, and Rule 32(c)(1) are unconstitutional as applied to Booker. Under their authority, the judge, rather than the jury, found the facts necessary to increase Booker's offense level pursuant to the listed provisions; the judge found those facts by a preponderance of the evidence, rather than beyond a reasonable doubt; and, on the basis of these findings, the judge imposed a sentence above the maximum legally permitted by the jury's findings. Thus, in Booker's case, the concerted action of § 3553(b)(1) and the operative Guidelines and the relevant Rule of Criminal Procedure resulted in unconstitutional judicial factfinding. The majority cannot pinpoint § 3553(b)(1) alone as the source of the violation.

At the same time, the majority's remedy is far too broad. We have before us only a single unconstitutional application of § 3553(b)(1) (and accompanying parts of the sentencing scheme). In such a case, facial invalidation is unprecedented. It is particularly inappropriate here, where it is evident that § 3553(b)(1) is entirely constitutional in numerous other applications. Fanfan's case is an example: The judge applied the Guidelines to the extent supported by the jury's findings. This application of § 3553(b)(1) was constitutional. To take another example, when the Government seeks a sentence within the Guidelines range supported by the jury's verdict, applying § 3553(b)(1) to restrict the judge's discretion to that Guidelines range is constitutional.

Section 3553(b)(1) is also constitutional when the Government seeks a sentence above the Guidelines range supported by the jury's verdict, but proves the facts supporting the enhancements to a jury beyond a reasonable doubt. Section 3553(b)(1) provides that "the court shall impose a sentence of the kind, and within the range," set by the Guidelines. (Emphasis added.) It says nothing, however, about the procedures the court must employ to determine the sentence it ultimately "impose[s]." It says nothing about whether, before imposing a sentence, the court may submit sentence-enhancing facts to the jury; and it says nothing about the standard of proof. Because it does not address at all the procedures for Guidelines sentencing proceedings, § 3553(b)(1) comfortably accommodates cases in which a court determines a defendant's Guidelines range by way of jury factfinding or admissions rather than judicial factfinding.

The Constitution does not prohibit what § 3553(b)(1) accomplishes—binding district courts to the Guidelines. It prohibits allowing a judge alone to make a finding that raises the sentence beyond the sentence that could have lawfully been imposed by reference to facts found by the jury or admitted by the defendant. Many applications of § 3553(b)(1) suffer from no such vice. Yet the majority, by facially invalidating the statute, also invalidates these unobjectionable applications of the statute and thereby ignores the longstanding distinction between as-applied and facial challenges.

Just as there is no reason to strike § 3553(b)(1) on its face, there is likewise no basis for striking any Guideline at issue here on its face. Respondents have not established that USSG § 1B1.3(a)(2), § 2D1.1(c)(2), § 3C1.1, or § 1B1.11(b)(2) is invalid in all its applications, as Salerno requires. To the contrary, numerous applications of these provisions are valid. Such applications include cases in which the defendant admits the relevant facts or the jury finds the relevant facts beyond a reasonable doubt. Like § 3553(b)(1), USSG §§ 1B1.3(a)(2), 2D1.1(c)(2), 3C1.1, and 1B1.11(b)(2) say nothing about who must find the facts supporting enhancements, or what standard of proof the prosecution must satisfy. They simply attach effects to certain facts; they do not prescribe procedures for determining those facts. Even § 1B1.1, which provides instructions for applying the Guidelines, directs an order in which the various provisions are to be applied ("[d]etermine the base offense level," § 1B1.1(b), then "[a]pply the adjustments," § 1B1.1(c)), but says nothing about the specific procedures a sentencing court may employ in determining the base offense level and applying adjustments.

Moreover, there is no basis for facially invalidating § 6A1.3 or Rule 32(c)(1). To be sure, § 6A1.3(b) and Rule 32(c)(1) prescribe procedure: They require the judge, acting alone, to resolve factual disputes. When Booker was sentenced, § 6A1.3(b) provided that "[t]he court shall resolve disputed sentencing factors at a sentencing hearing in accordance with Rule 32(c)(1), Fed. R. Crim. P." At the time, the relevant portions of Rule 32(c)(1) provided:

"At the sentencing hearing, the court ... must rule on any unresolved objections to the presentence report.... For each matter controverted, the court must make either a finding on the allegation or a determination that no finding is necessary because the controverted matter will not be taken into account in, or will not affect, sentencing." (Emphasis added.)

The natural meaning of "the court . . . must rule" is that the judge, without the jury, must resolve factual disputes as necessary. This Rule of Criminal Procedure, as applied at Booker's sentencing hearing, required the judge to make findings that increased Booker's offense level beyond the Guidelines range authorized by the jury. The application of the Rule to Booker therefore was unconstitutional.

Nonetheless, the Rule has other valid applications. For example, the Rule is valid when it requires the sentencing judge, without a jury, to resolve a factual dispute in order to decide where within the jury-authorized Guidelines range a defendant should be sentenced. The Rule is equally valid when it requires the judge to resolve a factual dispute in order to support a downward adjustment to the defendant's offense level.^[6]

Given the significant number of valid applications of all portions of the current sentencing scheme, we should not facially invalidate any particular section of the Federal Rules of Criminal Procedure, the Guidelines, or the SRA. Instead, we should invalidate only the application to Booker, at his previous sentencing hearing, of § 3553(b)(1); USSG §§ 1B1.3(a)(2), 2D1.1(c)(2), 3C1.1, 1B1.1, 1B1.11(b)(2), and 6A1.3(b); and Rule 32(c)(1).

II

Invalidating § 3553(b)(1), the Guidelines listed above, and Rule 32(c)(1) as applied to Booker by the District Court leaves the question whether the scheme's unconstitutional application to Booker can be severed from the scheme's many other constitutional applications to defendants like Fanfan. Severability doctrine is grounded in a presumption that Congress intends statutes to have effect to the full extent the Constitution allows.^[7]Regan v. Time, Inc., 468 U.S. 641, 652 (1984); Vermeule, Saving Constructions, 85 Geo. L. J. 1945, 1959-1963 (1997) (hereinafter Vermeule). The severability issue may arise when a court strikes either a provision of a statute or an application of a provision. Severability of provisions is perhaps more visible than severability of applications in our case law. See, e.g., Alaska Airlines, Inc. v. Brock, 480 U.S. 678, 684-697 (1987) (severing unconstitutional legislative veto provision from other provisions).^[8]

However, severability questions arise from unconstitutional applications of statutes as well. Congress often expressly provides for severance of unconstitutional applications.^[9] This Court has acknowledged the severability of applications in striking down some applications of a statute while leaving others standing. In Brockett, 472 U. S., at 504-507, the Court invalidated a state moral nuisance statute only insofar as it reached constitutionally protected materials, relying on the statute's severability clause. And in Tennessee v. Garner, 471 U.S. 1, 4 (1985), the Court considered a state statute that authorized police to use "`all the necessary means to effect [an] arrest.'" The Court held the statute unconstitutional insofar as it allowed the use of deadly force against an unarmed, nondangerous suspect; but it declined to invalidate the statute on its face, specifically noting that the statute could be applied constitutionally in other circumstances. Id., at 11-12. In Brockett and Garner, then, the Court recognized that the unconstitutional applications of the statutes were severable from the constitutional applications. The Court fashioned the remedy narrowly, in keeping with the usual presumption of severability.

I thus disagree with JUSTICE STEVENS that severability analysis does not apply. Ante, at 280-281, and n. 6 (opinion dissenting in part).^[10] I acknowledge that, as a general matter, the Court often disposes of as-applied challenges to a statute by simply invalidating particular applications of the statute, without saying anything at all about severability. See United States v. Grace, 461 U.S. 171, 183 (1983) (concluding that statute that prohibited carrying banners in the United States Supreme Court Building and on its grounds was unconstitutional as applied to the sidewalks surrounding the building); Edenfield v. Fane, 507 U.S. 761, 763 (1993) (striking down a solicitation ban on certified public accountants as applied "in the business context"); Treasury Employees, 513 U. S., at 501-503 (REHNQUIST, C. J., joined by SCALIA and THOMAS, JJ., dissenting) (expressing view that injunction against honoraria ban should be tailored to unconstitutional applications).

Such decisions (in which the Court is silent as to applications not before it) might be viewed as having conducted an implicit severability analysis. See id., at 485-489 (O'CONNOR, J., concurring in judgment in part and dissenting in part). A better view is that the parties in those cases could have raised the issue of severability, but did not bother, because (as is often the case) there was no arguable reason to defeat the presumption of severability. The unconstitutional applications of the statute were fully independent of and severable from the remaining constitutional applications. Here, the question is squarely presented: The parties press it, and there is extraordinary reason to clarify the remedy, namely, that our decision potentially affects every sentencing by the federal courts.

I therefore proceed to the severability question—whether the unconstitutional application of § 3553(b)(1); USSG §§ 1B1.3, 2D1.1(c)(2), 3C1.1, 1B1.1, 1B1.11(b)(2), and 6A1.3; and Rule 32(c)(1) to Booker is severable from the constitutional applications of these provisions. That is, even though we have invalidated the application of these provisions to Booker, may other defendants be sentenced pursuant to them? We presume that the unconstitutional application is severable. See, e.g., Regan, 468 U. S., at 653. This presumption is a manifestation of Salerno's general rule that we should not strike a statute on its face unless it is invalid in all its applications. Unless the Legislature clearly would not have enacted the constitutional applications independently of the unconstitutional application, the Court leaves the constitutional applications standing. 468 U. S., at 653.

Here, the presumption of severability has not been overcome. In light of the significant number of constitutional applications of the scheme, it is far from clear that Congress would not have passed the SRA or allowed Rule 32 to take effect, or that the Commission would not have promulgated the particular Guidelines at issue, had either body known that the application of the scheme to Booker was unconstitutional. Ante, at 274-279 (STEVENS, J., dissenting in part). As noted above, many applications of the Guidelines are constitutional: The defendant may admit the necessary facts; the Government may not seek enhancements beyond the offense level supported by the jury's verdict; the judge may find facts supporting an enhancement but (taking advantage of the overlap in Guidelines ranges) sentence the defendant within the jury-authorized range; or the jury may find the necessary facts.

Certainly it is not obvious that Congress would have preferred the entirely discretionary system that the majority fashions. The text and structure of the SRA show that Congress meant the Guidelines to bind judges. One of the purposes of the Commission, as set forth in the SRA, was to

"provide certainty and fairness in meeting the purposes of sentencing, avoiding unwarranted sentencing disparities among defendants with similar records who have been found guilty of similar criminal conduct while maintaining sufficient flexibility to permit individualized sentences when warranted by mitigating or aggravating factors not taken into account in the establishment of general sentencing practices." 28 U. S. C. § 991(b)(1)(B) (emphases added).

Accordingly, Congress made the Guidelines mandatory and closely circumscribed courts' authority to depart from the Guidelines range. 18 U. S. C. A. § 3553(b)(1) (Supp. 2004). Congress also limited appellate review of sentences imposed pursuant to the Guidelines to instances in which the sentence was (1) in violation of law, (2) a result of an incorrect application of the Guidelines, (3) outside the applicable Guidelines range, or (4) in the absence of an applicable Guideline, plainly unreasonable. § 3742(e) (main ed. and Supp. 2004). Striking down § 3553(b)(1) and the Guidelines only as applied to Booker (and other defendants who have received unconstitutional enhancements) would leave in place the essential framework of the mandatory system Congress created. Applying the Guidelines in a constitutional fashion affords some uniformity; total discretion, none. To suggest, as JUSTICE BREYER does, that a discretionary system would do otherwise, ante, at 249-253, 264 (opinion of the Court), either supposes that the system is discretionary in name only or overlooks the very nature of discretion. Either assumption is implausible.

The majority says that retaining the SRA and the Guidelines "engraft[s]" a jury trial requirement onto the sentencing scheme. Ante, at 246 (opinion of BREYER, J.). I am, of course, aware that, though severability analysis may proceed "by striking out or disregarding words [or, here, applications] that are in the [challenged] section," it may not proceed "by inserting [applications] that are not now there"; that would constitute legislation beyond our judicial power. United States v. Reese, 92 U. S. 214, 221 (1876). By allowing jury factfinding in some cases, however, we are no more "engrafting" a new requirement onto the statute than we do every time we invalidate a statute in some of the applications that the statute, on its face, appears to authorize. See, e. g., Brockett v. Spokane Arcades, Inc., 472 U. S. 491 (1985). I therefore do not find the "engraftment" label helpful as a means of judging the correctness of our severability analysis.

Granted, part of the severability inquiry is "whether the statute [as severed] will function in a manner consistent with the intent of Congress." Alaska Airlines, Inc., 480 U. S., at 685. Applying the Guidelines constitutionally (for instance, when admissions or jury findings support all upward enhancements) might seem at first glance to violate this principle. But so would the Government's proposal of applying the Guidelines as a whole to some defendants, but not others. The Court's solution violates it even more clearly by creating a system that eliminates the mandatory nature of the Guidelines. In the end, nothing except the Guidelines as written will function in a manner perfectly consistent with the intent of Congress, and the Guidelines as written are unconstitutional in some applications. While all of the remedial possibilities are thus, in a sense, second best, the solution JUSTICE STEVENS and I would adopt does the least violence to the statutory and regulatory scheme.

* * *

I would hold that § 3553(b)(1), the provisions of the Guidelines discussed above, and Rule 32(c)(1) are unconstitutional as applied to Booker, but that the Government has not overcome the presumption of severability. Accordingly, the unconstitutional application of the scheme in Booker's case is severable from the constitutional applications of the same scheme to other defendants. I respectfully dissent from the Court's contrary conclusion.

JUSTICE BREYER, with whom THE CHIEF JUSTICE, JUSTICE O'CONNOR, and JUSTICE KENNEDY join, dissenting in part.

The Court today applies its decisions in Apprendi v. New Jersey, 530 U. S. 466 (2000), and Blakely v. Washington, 542 U. S. 296 (2004), to the Federal Sentencing Guidelines. The Court holds that the Sixth Amendment requires a jury, not a judge, to find sentencing facts — facts about the way in which an offender committed the crime — where those facts would move an offender from lower to higher Guidelines ranges. I disagree with the Court's conclusion. I find nothing in the Sixth Amendment that forbids a sentencing judge to determine (as judges at sentencing have traditionally determined) the manner or way in which the offender carried out the crime of which he was convicted.

The Court's substantive holding rests upon its decisions in Apprendi, supra, and Blakely, supra. In Apprendi, the Court held that the Sixth Amendment requires juries to find beyond a reasonable doubt the existence of "any fact that increases the penalty for a crime" beyond "the prescribed statutory maximum." 530 U. S., at 490 (emphasis added). In Blakely, the Court defined the latter term as "the maximum sentence a judge may impose solely on the basis of the facts reflected in the jury verdict or admitted by the defendant." 542 U. S., at 303 (emphasis in original). Today, the Court applies its Blakely definition to the Federal Sentencing Guidelines. I continue to disagree with the constitutional analysis the Court set forth in Apprendi and in Blakely. But even were I to accept that analysis as valid, I would disagree with the way in which the Court applies it here.

I

THE CHIEF JUSTICE, JUSTICE O'CONNOR, JUSTICE KENNEDY, and I have previously explained at length why we cannot accept the Court's constitutional analysis. See Blakely, 542 U. S., at 314-326 (O'CONNOR, J., dissenting); id., at 326-328 (KENNEDY, J., dissenting); id., at 328-347 (BREYER, J., dissenting); Harris v. United States, 536 U. S. 545, 549-550 (2002) (KENNEDY, J., opinion of the Court); id., at 569-572 (BREYER, J., concurring in part and concurring in judgment); Apprendi, 530 U. S., at 523-554 (O'CONNOR, J., dissenting); id., at 555-556 (BREYER, J., dissenting); Jones v. United States, 526 U. S. 227, 264-272 (1999) (KENNEDY, J., dissenting); Monge v. California, 524 U. S. 721, 728-729 (1998) (O'CONNOR, J., opinion of the Court); McMillan v. Pennsylvania, 477 U. S. 79, 86-91 (1986) (REHNQUIST, C. J., opinion of the Court).

For one thing, we have found the Court's historical argument unpersuasive. See Blakely, supra, at 323 (O'CONNOR, J., dissenting); Apprendi, supra, at 525-528 (O'CONNOR, J., dissenting). Indeed, the Court's opinion today illustrates the historical mistake upon which its conclusions rest. The Court reiterates its view that the right of "`trial by jury has been understood to require'" a jury trial for determination of "`the truth of every accusation.'" Ante, at 239 (opinion of STEVENS, J.) (quoting Apprendi, supra, at 477; emphasis in original). This claim makes historical sense insofar as an "accusation" encompasses each factual element of the crime of which a defendant is accused. See, e. g., United States v. Gaudin, 515 U. S. 506, 509-510, 522-523 (1995). But the key question here is whether that word also encompasses sentencing facts — facts about the offender (say, recidivism) or about the way in which the offender committed the crime (say, the seriousness of the injury or the amount stolen) that help a sentencing judge determine a convicted offender's specific sentence.

History does not support a "right to jury trial" in respect to sentencing facts. Traditionally, the law has distinguished between facts that are elements of crimes and facts that are relevant only to sentencing. See, e. g., Almendarez-Torres v. United States, 523 U. S. 224, 228 (1998); Witte v. United States, 515 U. S. 389, 399 (1995); United States v. Watts, 519 U. S. 148, 154 (1997) (per curiam); United States v. Dunnigan, 507 U. S. 87, 97 (1993); Mistretta v. United States, 488 U. S. 361, 396 (1989). Traditionally, federal law has looked to judges, not to juries, to resolve disputes about sentencing facts. See, e. g., Fed. Rule Crim. Proc. 32(a). Traditionally, those familiar with the criminal justice system have found separate, postconviction judge-run sentencing procedures sensible given the difficulty of obtaining relevant sentencing information before the moment of conviction. They have found those proceedings practical given the impracticality of the alternatives, say, two-stage (guilt, sentence) jury procedures. See, e. g., Judicial Conference of the United States, Committee on Defender Services, Subcommittee on Federal Death Penalty Cases, Federal Death Penalty Cases: Recommendations Concerning the Cost and Quality of Defense Representation 9-10 (May 1998). And, despite the absence of jury determinations, they have found those proceedings fair as long as the convicted offender has the opportunity to contest a claimed fact before the judge, and as long as the sentence falls within the maximum of the range that a congressional statute specifically sets forth.

The administrative rules at issue here, Federal Sentencing Guidelines, focus on sentencing facts. They circumscribe a federal judge's sentencing discretion in respect to such facts, but in doing so, they do not change the nature of those facts. The sentencing courts continue to use those facts, not to convict a person of a crime as a statute defines it, but to help determine an appropriate punishment. Thus, the Court cannot ground today's holding in a "constitutional tradition assimilated from the common law" or in "the Magna Carta." Ante, at 238-239 (opinion of STEVENS, J.). It cannot look to the Framers for support, for they, too, enacted criminal statutes with indeterminate sentences, revealing their own understanding and acceptance of the judge's factfinding role at sentencing. See Act of Apr. 30, 1790, ch. 9, 1 Stat. 112-118.

Indeed, it is difficult for the Court to find historical support other than in two recent cases, Apprendi and Blakely — cases that we, like lower courts, read not as confirming, but as confounding a pre-Apprendi, pre-Blakely legal tradition that stretches back a century or more. See, e. g., Williams v. New York, 337 U. S. 241, 246 (1949); cf., e. g., 375 F. 3d 508, 514 (CA7 2004) (case below) ("Blakely redefined `statutory maximum'"); United States v. Ameline, 376 F. 3d 967, 973 (CA9 2004) ("Blakely court worked a sea change in the body of sentencing law"); United States v. Pineiro, 377 F. 3d 464, 468-469 (CA5 2004) (same); see also United States v. Penaranda, 375 F. 3d 238, 243, n. 5 (CA2 2004) (same, collecting cases).

For another thing, applied in the federal context of mandatory Guidelines, the Court's Sixth Amendment decision would risk unwieldy trials, a two-tier jury system, a return to judicial sentencing discretion, or the replacement of sentencing ranges with specific mandatory sentences. Cf. Blakely, 542 U. S., at 330-340 (BREYER, J., dissenting). The decision would pose a serious obstacle to congressional efforts to create a sentencing law that would mandate more similar treatment of like offenders, that would thereby diminish sentencing disparity, and that would consequently help to overcome irrational discrimination (including racial discrimination) in sentencing. See id., at 315-316 (O'CONNOR, J., dissenting). These consequences would seem perverse when viewed through the lens of a Constitution that seeks a fair criminal process.

The upshot is that the Court's Sixth Amendment decisions — Apprendi, Blakely, and today's — deprive Congress and state legislatures of authority that is constitutionally theirs. Cf. Blakely, supra, at 326-328 (KENNEDY, J., dissenting); Apprendi, 530 U. S., at 544-545 (O'CONNOR, J., dissenting); id., at 560-564 (BREYER, J., dissenting). The "sentencing function long has been a peculiarly shared responsibility among the Branches of Government." Mistretta, supra, at 390. Congress' share of this joint responsibility has long included not only the power to define crimes (by enacting statutes setting forth their factual elements) but also the power to specify sentences, whether by setting forth a range of individual-crime-related sentences (say, 0-to-10 years' imprisonment for bank robbery) or by identifying sentencing factors that permit or require a judge to impose higher or lower sentences in particular circumstances. See, e. g., Almendarez-Torres, supra, at 228; McMillan, 477 U. S., at 85.

This last mentioned power is not absolute. As the Court suggested in McMillan, confirmed in Almendarez-Torres, and recognized but rejected in Blakely, one might read the Sixth Amendment as permitting "legislatures" to "establish legally essential [judge-determined] sentencing factors within [say, due process] limits." Blakely, supra, at 307 (emphasis in original); cf. Almendarez-Torres, supra, at 228 (distinguishing between "elements" and "factors relevant only to ... sentencing," and noting that, "[w]ithin limits, the question of which factors are which is normally a matter for Congress" (citation omitted)); McMillan, supra, at 88 (upholding a Pennsylvania statute in part because it gave "no impression of having been tailored to permit the [sentencing factor] finding to be a tail which wags the dog of the substantive offense"). But the power does give Congress a degree of freedom (within constraints of fairness) to choose to characterize a fact as a "sentencing factor," relevant only to punishment, or as an element of a crime, relevant to guilt or innocence. The Court has rejected this approach apparently because it finds too difficult the judicial job of managing the "fairness" constraint, i. e., of determining when Congress has overreached. But the Court has nowhere asked, "compared to what?" Had it done so, it could not have found the practical difficulty it has mentioned, Blakely, supra, at 307-308, sufficient to justify the severe limits that its approach imposes upon Congress' legislative authority.

These considerations — of history, of constitutionally relevant consequences, and of constitutional authority — have been more fully discussed in other opinions. See, e. g., Blakely, 542 U. S., at 314-326 (O'CONNOR, J., dissenting); id., at 327-328 (KENNEDY, J., dissenting); id., at 328-347 (BREYER, J., dissenting); Harris, 536 U. S., at 549-550, 569-572; Apprendi, supra, at 523-554, 555-556; McMillan, supra, at 86-91. I need not elaborate them further.

II

Although the considerations just mentioned did not dissuade the Court from its holdings in Apprendi and Blakely, I should have hoped they would have dissuaded the Court from extending those holdings to the statute and Guidelines at issue here. See Sentencing Reform Act of 1984, as amended, 18 U. S. C. § 3551 et seq., 28 U. S. C. § 991 et seq.; United States Sentencing Commission, Guidelines Manual (Nov. 2003) (USSG). Legal logic does not require that extension, for there are key differences.

First, the Federal Guidelines are not statutes. The rules they set forth are administrative, not statutory, in nature. Members, not of Congress, but of a Judicial Branch Commission, wrote those rules. The rules do not "establis[h] minimum and maximum penalties" for individual crimes, but guide sentencing courts, only to a degree, "fetter[ing] the discretion of sentencing judges to do what they have done for generations — impose sentences within the broad limits established by Congress." Mistretta, 488 U. S., at 396; see also USSG § 5G1.1; cf. Witte, 515 U. S., at 399 (explaining that the Guidelines range "still falls within the scope of the legislatively authorized penalty"). The rules do not create a new set of legislatively determined sentences so much as they reflect, organize, rationalize, and modify an old set of judicially determined pre-Guidelines sentences. See 28 U. S. C. § 994(a); USSG § 1A1.1, editorial note, § 3, pp. 2-4 (describing the Commission's empirical approach). Thus, the rules do not, in Apprendi's words, set forth a "prescribed statutory maximum," 530 U. S., at 490 (emphasis added), as the law has traditionally understood that phrase.

I concede that Blakely defined "prescribed statutory maximum" more broadly as "the maximum sentence a judge may impose solely on the basis of the facts reflected in the jury verdict or admitted by the defendant." 542 U. S., at 303 (emphasis omitted). But the Court need not read this language as extending the scope of Apprendi. Blakely purports to follow, not to extend, Apprendi. 542 U. S., at 301. And Blakely, like Apprendi, involved sentences embodied in a statute, not in administrative rules.

More importantly, there is less justification for applying an Apprendi-type constitutional rule where administrative guidelines, not statutes, are at issue. The Court applies its constitutional rule to statutes in part to avoid what Blakely sees as a serious problem, namely, a legislature's ability to make of a particular fact an "element" of a crime or a sentencing factor, at will. See ante, at 230 (opinion of STEVENS, J.). That problem — that legislative temptation — is severely diminished when Commission Guidelines are at issue, for the Commission cannot create "elements" of crimes. It cannot write rules that "bind or regulate the primary conduct of the public." Mistretta, supra, at 396. Rather, it must write rules that reflect what the law has traditionally understood as sentencing factors. That is to say, the Commission cannot switch between "elements" and "sentencing factors" at will because it cannot write substantive criminal statutes at all. See 28 U. S. C. § 994(a); cf. Blakely, supra, at 301-302, 306-307.

At the same time, to extend Blakely's holding to administratively written sentencing rules risks added legal confusion and uncertainty. Read literally, Blakely's language would include within Apprendi's strictures a host of nonstatutory sentencing determinations, including appellate court decisions delineating the limits of the legally "reasonable." (Imagine an appellate opinion that says a sentence for ordinary robbery greater than five years is unreasonably long unless a special factor, such as possession of a gun, is present.) Indeed, read literally, Blakely's holding would apply to a single judge's determination of the factors that make a particular sentence disproportionate or proportionate. (Imagine a single judge setting forth, as a binding rule of law, the legal proposition about robbery sentences just mentioned.) Appellate courts' efforts to define the limits of the "reasonable" of course would fall outside Blakely's scope. But they would do so not because they escape Blakely's literal language, but because they are not legislative efforts to create limits. Neither are the Guidelines legislative efforts. See Mistretta, supra, at 412.

Second, the sentencing statutes at issue in Blakely imposed absolute constraints on a judge's sentencing discretion, while the federal sentencing statutes here at issue do not. As the Blakely Court emphasized, the Washington statutes authorized a higher-than-standard sentence on the basis of a factual finding only if the fact in question was a new fact — i. e., a fact that did not constitute an element of the crime of conviction or an element of any more serious or additional crime. 542 U. S., at 301-302, 306-307. A judge applying those statutes could not even consider, much less impose, an exceptional sentence, unless he found facts "`other than those which are used in computing the standard range sentence for the offense.'" Id., at 299 (quoting State v. Gore, 143 Wash. 2d 288, 315-316, 21 P. 3d 262, 277 (2001)).

The federal sentencing statutes, however, offer a defendant no such fact-related assurance. As long as "there exists an aggravating or mitigating circumstance of a kind, or to a degree, not adequately taken into consideration by the Sentencing Commission," 18 U. S. C. A. § 3553(b)(1) (Supp. 2004), they permit a judge to depart from a Guidelines sentence based on facts that constitute elements of the crime (say, a bank robbery involving a threat to use a weapon, where the weapon in question is nerve gas). Whether departure-triggering circumstances exist in a particular case is a matter for a court, not for Congress, to decide.

Thus, as far as the federal statutes are concerned, the federal system, unlike the state system at issue in Blakely, provides a defendant with no guarantee that the jury's finding of factual elements will result in a sentence lower than the statutory maximum. Rather, the statutes put a potential federal defendant on notice that a judge conceivably might sentence him anywhere within the range provided by statute — regardless of the applicable Guidelines range. See Witte, 515 U. S., at 399; see also Comment, Sixth Amendment — State Sentencing Guidelines, 118 Harv. L. Rev. 333, 339-340 (2004). Hence as a practical matter, they grant a potential federal defendant less assurance of a lower Guidelines sentence than did the state statutes at issue in Blakely.

These differences distinguish these cases from Apprendi and Blakely. They offer a principled basis for refusing to extend Apprendi's rule to these cases.

III

For these reasons, I respectfully dissent.

NOTES

[*] Together with No. 04-105, United States v. Fanfan, on certiorari before judgment to the United States Court of Appeals for the First Circuit.

[†] Briefs of amici curiae urging reversal in both cases were filed for the United States Sentencing Commission by James K. Robinson, Charles R. Tetzlaff, and Pamela O. Barron; and for the Honorable Orrin G. Hatch et al. by Gregory G. Garre.

Briefs of amici curiae urging affirmance in both cases were filed for Families Against Mandatory Minimums by Gregory L. Poe, Roy T. Englert, Jr., Max Huffman, and Mary Price; for the Federal Public Defender, Northern District of Texas, by Ira R. Kirkendoll and Carlos R. Cardona; for the National Association of Criminal Defense Lawyers by Samuel J. Buffone, David O. Stewart, Thomas C. Goldstein, Amy Howe, and David M. Porter; for the National Association of Federal Defenders by Paul M. Rashkind, Carol A. Brook, Henry J. Bemporad, and Frances H. Pratt; for the New York Council of Defense Lawyers by Alexandra A. E. Shapiro and Lewis J. Liman; for the Washington Legal Foundation et al. by Donald B. Verrilli, Jr., Elaine J. Goldenberg, Daniel J. Popeo, and Paul D. Kamenar; and for Thomas F. Liotti, by Mr. Liotti, pro se.

John S. Martin, Jr., filed a brief for an Ad Hoc Group of Former Federal Judges as amici curiae in both cases.

[*] JUSTICE SCALIA, JUSTICE SOUTER, JUSTICE THOMAS, and JUSTICE GINSBURG join this opinion.

[1] The questions presented are:

"1. Whether the Sixth Amendment is violated by the imposition of an enhanced sentence under the United States Sentencing Guidelines based on the sentencing judge's determination of a fact (other than a prior conviction) that was not found by the jury or admitted by the defendant.

"2. If the answer to the first question is `yes,' the following question is presented: whether, in a case in which the Guidelines would require the court to find a sentence-enhancing fact, the Sentencing Guidelines as a whole would be inapplicable, as a matter of severability analysis, such that the sentencing court must exercise its discretion to sentence the defendant within the maximum and minimum set by statute for the offense of conviction." E. g., Pet. for Cert. in No. 04-104, p. (I).

[2] In Mistretta v. United States, 488 U. S. 361 (1989), we pointed out that Congress chose explicitly to adopt a "mandatory-guideline system" rather than a system that would have been "only advisory," and that the statute "makes the Sentencing Commission's guidelines binding on the courts." Id., at 367.

[3] 18 U. S. C. A. § 3553(a) (main ed. and Supp. 2004).

[4] Watts, in particular, presented a very narrow question regarding the interaction of the Guidelines with the Double Jeopardy Clause, and did not even have the benefit of full briefing or oral argument. It is unsurprising that we failed to consider fully the issues presented to us in these cases. See 519 U. S., at 171 (KENNEDY, J., dissenting).

[5] We added: "Instead, petitioners argue that the judge might have made different factual findings if only the judge had known that the law required him to assume the jury had found a cocaine-only, not a cocaine-and-crack, conspiracy. It is sufficient for present purposes, however, to point out that petitioners did not make this particular argument in the District Court. Indeed, they seem to have raised their entire argument for the first time in the Court of Appeals. Thus, petitioners did not explain to the sentencing judge how their `jury-found-only-cocaine' assumption could have made a difference to the judge's own findings, nor did they explain how this assumption (given the judge's findings) should lead to greater leniency." Edwards, 523 U. S., at 515-516.

[*] THE CHIEF JUSTICE, JUSTICE O'CONNOR, JUSTICE KENNEDY, and JUSTICE GINSBURG join this opinion.

[1] We have, on occasion, debated the proper interpretation of various precedents concerning facial challenges to statutes. Compare Chicago v. Morales, 527 U. S. 41, 54-55, n. 22 (1999) (plurality opinion), with id., at 78-83 (SCALIA, J., dissenting), and United States v. Salerno, 481 U. S. 739, 745 (1987). That debate is immaterial to my conclusion here, because it borders on the frivolous to contend that the Guidelines can be constitutionally applied "only in a fraction of the cases [they were] originally designed to cover." United States v. Raines, 362 U. S. 17, 23 (1960).

[2] See also Lodging of Government, Estimate of Number of Cases Possibly Impacted by the Blakely Decision, p. 2 (hereinafter Estimate).

[3] See, e. g., Webster v. Reproductive Health Services, 492 U. S. 490, 524 (1989) (O'CONNOR, J., concurring in part and concurring in judgment) (arguing that a statute cannot be struck down on its face whenever the statute has "some quite straightforward applications [that] would be constitutional"); Secretary of State of Md. v. Joseph H. Munson Co., 467 U. S. 947, 977 (1984) (REHNQUIST, J., dissenting) ("When a litigant challenges the constitutionality of a statute, he challenges the statute's application to him.... If he prevails, the Court invalidates the statute, not in toto, but only as applied to those activities. The law is refined by preventing improper applications on a case-by-case basis. In the meantime, the interests underlying the law can still be served by its enforcement within constitutional bounds"); cf. Raines, 362 U. S., at 21 (this Court should never "`formulate a rule of constitutional law broader than is required by the precise facts to which it is to be applied'"); Ohio v. Akron Center for Reproductive Health, 497 U. S. 502, 514 (1990) (plurality opinion) (statutes should not be invalidated "on a facial challenge based upon a worst-case analysis that may never occur").

[4] Memorandum from Christopher A. Wray, Assistant Attorney General, U. S. Department of Justice, Criminal Division, to All Federal Prosecutors, re: Guidance Regarding the Application of Blakely v. Washington to Pending Cases, p. 8, available at http://sentencing.typepad.com/sentencing_law_and_policy/files/chris_wray_doj_memo.pdf (hereinafter Application of Blakely); see also Brief for National Association of Federal Defenders as Amicus Curiae 9-12.

[5] See Application of Blakely 9.

[6] There is a line of cases that some commentators have described as standing for the proposition that the Court must engage in severability analysis if a statute is unconstitutional in only some of its applications. See Stern 82. However, these cases simply hold that a statute that may apply both to situations within the scope of Congress' enumerated powers and also to situations that exceed such powers, the Court will sustain the statute only if it can be validly limited to the former situations, and will strike it down if it cannot be so limited. Compare United States v. Reese, 92 U.S. 214, 221 (1876) (invalidating in its entirety statute that punished individuals who interfered with the right to vote, when the statute applied to conduct that violated the Fifteenth Amendment and conduct outside that constitutional prohibition), and Trade-Mark Cases, 100 U.S. 82, 98 (1879) (concluding that the Trade-Mark Act must be read to "establish a universal system of trade-mark registration" and thus was invalid in its entirety because it exceeded the bounds of the Commerce Clause), with The Abby Dodge, 223 U.S. 166, 175 (1912) (construing language to apply only to waters not within the jurisdiction of the States, and therefore entirely valid), and NLRB v. Jones & Laughlin Steel Corp., 301 U.S. 1, 30-31 (1937) (holding that the National Labor Relations Act applied only to interstate commerce, and upholding its constitutionality on that basis). These cases are thus about constitutional avoidance, not severability.

In a separate dissent, JUSTICE THOMAS relies on this principle to conclude that the proper analysis is whether the unconstitutional applications of the Guidelines are sufficiently numerous and integral to warrant invalidating the Guidelines in their entirety. See post, at 323. While I understand the intuitive appeal of JUSTICE THOMAS' dissent, I do not believe that our cases support this approach. In any event, given the vast number of constitutional applications, see supra, at 276, it is clear that Congress would, as JUSTICE THOMAS concludes, prefer that the Guidelines not be invalidated. I therefore do not believe that any extension of our severability cases is warranted.

[7] The predicate for the Court's remedy is its assumption that Congress would not have enacted mandatory Guidelines if it had realized that the Sixth Amendment would require some enhancements to be supported by jury factfinding. If Congress should reenact the statute following our decision today, it would repudiate that premise. That is why I find the Court's professed disagreement with this proposition unpersuasive. See ante, at 250 (opinion of BREYER, J.). Surely Congress could reenact the identical substantive provisions if the reenactment included a clarifying provision stating that the word "court" shall not be construed to prohibit a judge from requiring jury factfinding when necessary to comply with the Sixth Amendment. Indeed, because in my view such a construction of the word "court" is appropriate in any event, see infra, at 286-287, there would be no need to include the clarifying provision to save the statute.

[8] In Sloan v. Lemon, 413 U. S. 825 (1973), the Court concluded that legislation reimbursing parents for tuition paid to private schools ran afoul of the Establishment Clause and struck down the law in its entirety, even as applied to parents of students in secular schools. The Court did not, as the majority would have us do, strike down particular parts of the statute. In Welsh v. United States, 398 U. S. 333, 361-363 (1970), Justice Harlan, writing alone, concluded that a statutory provision that allowed conscientious objectors to be exempt from military service only if their views were religiously based violated the Establishment Clause. He then concluded that, rather than deny the exception to religiously based objectors, it should be extended to moral objectors, in large part because "the broad discretion conferred by a severability clause" was not present in the case. Id., at 365. Finally, in Heckler v. Mathews, 465 U. S. 728, 739, n. 6 (1984), the Court stated the obvious rule that when a statute provides a benefit to one protected class and not the other, the Court is faced with the choice of requiring the Legislature to extend the benefits, or nullifying the benefits altogether. None of these cases stands for the sweeping proposition that where parts of a statute are invalid in certain applications, the Court may opine as to whether Congress would prefer facial invalidation of some, but not all, of the provisions necessary to the constitutional violation.

[9] The Commissioners went on to note that, "[e]ven if Blakely is found to apply to the federal guidelines, the waters are not as choppy as some would make them out to be. The viability of the [Guidelines] previously was called into question by some after [Apprendi v. New Jersey, 530 U. S. 466 (2000)]. After an initial period of uncertainty, however, the circuit courts issued opinions and the Department of Justice instituted procedures to ensure that future cases complied with Apprendi's requirements and also left the guidelines system intact." Hearings on Blakely 1.

[10] This argument finds support in the Government's successful adaptation to our decision in Apprendi. After that decision, prosecutors began to allege more and more "sentencing factors" in indictments. See supra, at 277. The Government's ability to do so suggests that the Guidelines are far more compatible with "jury factfinding" than the Court admits. And, the fact that Congress is presumably aware of the Government's practices in light of Apprendi, yet has not condemned the practices or taken any actions to reform them, indicates that limited jury factfinding is, contrary to the majority's assertion, compatible with legislative intent. See ante, at 250 (opinion of BREYER, J.).

[11] See M. Johnson & S. Gilbert, The U. S. Sentencing Guidelines: Results of the Federal Judicial Center's 1996 Survey 7-9 (1997) (noting that among federal judges and probation officers, there is widespread "frustration with the power and discretion held by prosecutors under the guidelines" and that "guidelines are manipulated through plea agreements"); Saris, Have the Sentencing Guidelines Eliminated Disparity? One Judge's Perspective, 30 Suffolk U. L. Rev. 1027, 1030 (1997); see also Nagel & Schulhofer, A Tale of Three Cities: An Empirical Study of Charging and Bargaining Practices Under the Federal Sentencing Guidelines, 66 S. Cal. L. Rev. 501, 560 (1992) (arguing that fact bargaining is common under the Guidelines and has resulted in substantial sentencing disparities).

[12] Incidentally, the original version of S. 1437 looked much like the regime that the Court has mandated today—it directed the sentencing judge to consider a variety of factors, only one of which was the sentencing range established by the Guidelines, and subjected the ultimately chosen sentence to appellate review under a "clearly unreasonable" standard. See S. 1437, § 101 (proposed 18 U. S. C. §§ 2003(a), 3725(e)). That law was amended twice before it passed, the first time to include a mandatory directive to trial judges to impose a sentence within the Guidelines range, and the second time to change the standard of review from "`clearly unreasonable" to "`unreasonable.'" See Stith & Koh 245 (detailing amendments to S. 1437 prior to passage). It is worth noting that Congress had countless opportunities over the course of seven years of debate to enact the law the Court creates today. Congress' repeated rejection of proposed legislation constitutes powerful evidence that Congress did not want it to become law.

[13] See Stith & Koh 269-270; see also Wilkins, Newton, & Steer, Competing Sentencing Policies in a "War on Drugs" Era, 28 Wake Forest L. Rev. 305, 313 (1993) (same).

[14] See, e.g., 133 Cong. Rec. 33109 (1987) (remarks of Sen. Hatch) ("[T]he core function of the guidelines and the underlying statute ... is to reduce disparity in sentencing and restore fairness and predictability to the sentencing process. Adherence to the guidelines is therefore properly required under the law except in . . . rare and particularly unusual instances . . ."); id., at 33110 (remarks of Sen. Biden) ("That notion of allowing the courts to, in effect, second-guess the wisdom of any sentencing guideline is plainly contrary to the act's purpose of having a sentencing guidelines system that is mandatory, except when the court finds a circumstance meeting the standard articulated in § 3553(b). It is also contrary to the purpose of having Congress, rather than the courts, review the sentencing guidelines for the appropriateness of authorized levels of punishment"); S. Rep. No. 98-223, p. 76 (1983) (noting that the Senate Judiciary Committee "resisted [the] attempt to make the sentencing guidelines more voluntary than mandatory, because of the poor record of States reported in the National Academy of Science Report which have experimented with `voluntary' guidelines"); id., at 34-35 (citing the "urgent need for" sentencing reform because of sentencing disparities caused "directly [by] the unfettered discretion the law confers on [sentencing] judges and parole authorities responsible for imposing and implementing the sentence"); id., at 36-43, 62 (cataloging the "astounding" variations in federal sentencing and criticizing the unfairness of sentencing disparities).

[15] Indeed, the Court's contention that real conduct sentencing was the principal aim of the SRA finds no support in the legislative history. The only authority the Court cites is 18 U.S.C. § 3661, which permits a judge to consider any information she considers relevant to sentencing. See ante, at 249-250 (opinion of BREYER, J.). That provision, however, was enacted in 1970, see Pub. L. 91-452, § 1001(a), 84 Stat. 951 (there numbered § 3577), and thus provides no evidence whatsoever of Congress' intent when it passed the SRA in 1984. Clearly, Congress thought that real conduct sentencing could not effectively address sentencing disparities without a binding Guidelines regime. For this reason, traditional sentencing goals have always played a minor role in the Guidelines system: "While the thick-as-your-wrist Guideline Manual specifically directs sentencing judges to make thousands of determinations on discrete points, not once does it expressly direct that a specific decision leading to the applicable guideline range on the 256-box grid should or must turn on an individualized consideration of the traditional goals of sentencing." Osler, Uniformity and Traditional Sentencing Goals in the Age of Feeney, 16 Fed. Sentencing Rptr. 253, 253-254 (2004).

[16] Although there was no accompanying committee report attached to the PROTECT Act, the floor debates over the Act's relevant provisions belie the majority's contention that a discretionary Guidelines system is more consistent with Congress' intent than the holding I would adopt. See 149 Cong. Rec. S5113, S5121-S5122 (Apr. 10, 2003) (remarks of Sen. Hatch) (arguing that the PROTECT Act "says the game is over for judges: You will have some departure guidelines from the Sentencing Commission, but you are not going to go beyond those, and you are not going to go on doing what is happening in our society today on children's crimes, no matter how softhearted you are. That is what we are trying to do here.... We say in this bill: We are sick of this, judges. You are not going to do this anymore except within the guidelines set by the Sentencing Commission"); id., at S5123 ("[T]rial judges systematically undermine the sentencing guidelines by creating new reasons to reduce these sentences"); id., at S6708, S6711 (May 20, 2003) (remarks of Sen. Kennedy) ("The Feeney Amendment effectively strips Federal judges of discretion to impose individualized sentences, and transforms the longstanding sentencing guidelines system into a mandatory minimum sentencing system. It limits in several ways the ability of judges to depart downwards from the guidelines").

[17] Moreover, even if the change to an indeterminate system were necessary, the Court could have minimized the consequences to the system by limiting the application of its holding to those defendants on direct review who actually suffered a Sixth Amendment violation. Griffith v. Kentucky, 479 U. S. 314 (1987), does not require blind application of every part of this Court's holdings to all pending cases, but rather, requires that we apply any new "rule to all similar cases pending on direct review." Id., at 323. For obvious reasons, not all pending cases are made similar to Booker's and Fanfan's merely because they involved an application of the Guidelines.

[1] Part III of JUSTICE STEVENS's dissent relies in large part on legislative history. I agree with his assertion that "[t]he text of the law that actually passed Congress ... should be more than sufficient to demonstrate Congress' unmistakable commitment to a binding Guidelines system." Ante, at 294. I would not resort to committee reports and statements by various individuals, none of which constitutes action taken or interpretations adopted by Congress. "One determines what Congress would have done by examining what it did." Legal Services Corporation v. Velazquez, 531 U. S. 533, 560 (2001) (SCALIA, J., dissenting).

[2] See, e. g., ante, at 246 (noting that Congress intended the Guidelines system to achieve "increased uniformity of sentencing"); ante, at 250 (referring to "diminish[ing] sentencing disparity" as "Congress' basic statutory goal"); ante, at 255 ("Congress enacted the sentencing statutes in major part to achieve greater uniformity in sentencing"); ante, at 267 (referring to "Congress' basic objective of promoting uniformity in sentencing"); see also United States Sentencing Commission, Fifteen Years of Guidelines Sentencing xvi (Nov. 2004) ("Sentencing reform has had its greatest impact controlling disparity arising from the source at which the guidelines themselves were targeted—judicial discretion"); id., at 140 ("[T]he guidelines have succeeded at the job they were principally designed to do: reduce unwarranted disparity arising from differences among judges").

[3] Although the Guidelines took pre-existing sentencing practices into account, they are the product of policy decisions by the Sentencing Commission — including, for instance, decisions to call for sentences "significantly more severe than past practice" for the "most frequently sentenced offenses in the federal courts." Id., at 47. If those policy decisions are no longer mandatory, the sentencing judge is free to disagree with them.

[4] The closest the remedial majority dares come to an assertion that the Guidelines must be followed is the carefully crafted statement that "[t]he district courts, while not bound to apply the Guidelines, must consult those Guidelines and take them into account when sentencing." Ante, at 264. The remedial majority also notes that the Guidelines represent what the Sentencing Commission "finds to be better sentencing practices." Ante, at 263. True enough, but the Commission's view of what is "better" is no longer authoritative, and district judges are free to disagree — as are appellate judges.

[5] Paragraph (e)(1) requires a court of appeals to determine whether a sentence "was imposed in violation of law." 18 U.S.C. § 3742. Courts of appeals had of course always done this.

[6] In the face of this immense reality, it is almost captious to point out that some of the text of the preserved subsection (f) plainly assumes the binding nature of the Guidelines — for example, the reference to a "sentence ... imposed as a result of an incorrect application of the sentencing guidelines," § 3742(f)(1) (Supp. 2004), and the reference to a "departure . . . based on an impermissible factor," § 3742(f)(2). Moreover, paragraph (f)(1) requires the appellate court to "remand . . . for further sentencing proceedings" any case in which the sentence was imposed "as a result of an incorrect application of the sentencing guidelines." It is incomprehensible how or why this instruction can be combined with an obligation upon the appellate court to conduct its own independent evaluation of the "reasonableness" of a sentence.

[7] The four kinds of appeal arise when, respectively,

(1) the sentence is "imposed in violation of law," §§ 3742(a)(1), (b)(1), (e)(1), (f)(1) (main ed. and Supp. 2004);

(2) the sentence is "imposed as a result of an incorrect application of the sentencing guidelines," §§ 3742(a)(2), (b)(2), (e)(2), (f)(1);

(3) the sentence is either above or below "the applicable guideline range," §§ 3742(a)(3), (b)(3), (e)(3), (f)(2); and

(4) no guideline is applicable and the sentence is "plainly unreasonable," §§ 3742(a)(4), (b)(4), (e)(4), (f)(2).

[8] Pierce v. Underwood, 487 U.S. 552, 558-560 (1988), Cooter & Gell v. Hartmarx Corp., 496 U.S. 384, 403-405 (1990), and Koon v. United States, 518 U.S. 81, 99 (1996).

[9] Deciding whether a departure from a mandatory sentence (for a reason not taken into account in the Guidelines) is "unreasonable" (as § 3742(e)(3) required), or whether a sentence imposed for one of the rare offenses not covered by the Guidelines—though surrounded by mandatory sentences for related and analogous offenses—is "plainly unreasonable" (as § 3742(e)(4) required), differs toto caelo from determining, in the absence of any mandatory scheme, that a particular sentence is "unreasonable."

[1] Because the majority invalidates 18 U.S.C.A. § 3553(b)(1) (Supp. 2004) on its face, it is driven also to invalidate § 3742(e) (main ed. and Supp. 2004), which establishes standards of review for sentences and is premised on the binding nature of the Guidelines. See, e.g., § 3742(e)(2) (main ed.) (directing the court of appeals to determine whether the sentence "was imposed as a result of an incorrect application of the sentencing guidelines"); § 3742(e)(3) (Supp. 2004) (directing the court of appeals to determine whether the sentence "is outside the applicable guideline range" and satisfies other factors). Given that (as I explain) there is no warrant for striking § 3553(b)(1) on its face, striking § 3742(e) as well only does further needless violence to the statutory scheme.

[2] Booker's base offense level (supported by the facts the jury found) was 32. See United States Sentencing Commission, Guidelines Manual § 2D1.1(c)(4) (Nov. 2003) (USSG) (setting the base offense level for the crime of possession with intent to sell 50 to 150 grams of cocaine base at 32).

[3] The District Court applied the version of the Guidelines effective November 1, 2003.

[4] I take no position on whether USSG § 6A1.3, a policy statement, bound the District Court. Cf. Stinson v. United States, 508 U.S. 36, 42-43 (1993); Williams v. United States, 503 U.S. 193, 200-201 (1992). In any case, Rule 32(c)(1), which had the same effect as § 6A1.3, certainly bound the court.

[5] In 2002, Rule 32(c)(1) was amended and replaced with Rule 32(i)(3). The new Rule provides, in substantially similar fashion, that at sentencing, the court "must—for any disputed portion of the presentence report or other controverted matter—rule on the dispute or determine that a ruling is unnecessary either because the matter will not affect sentencing, or because the court will not consider the matter in sentencing." Fed. Rule Crim. Proc. 32(i)(3)(B) (2003).

[6] The commentary to § 6A1.3 states that "[t]he Commission believes that use of a preponderance of the evidence standard is appropriate to meet due process requirements and policy concerns in resolving disputes regarding application of the guidelines to the facts of a case." The Court's holding today corrects this mistaken belief. The Fifth Amendment requires proof beyond a reasonable doubt, not by a preponderance of the evidence, of any fact that increases the sentence beyond what could have been lawfully imposed on the basis of facts found by the jury or admitted by the defendant.

[7] I assume, without deciding, that our severability precedents—which require a nebulous inquiry into hypothetical congressional intent—are valid, a point the parties do not contest. I also assume that our doctrine on severability and facial challenges applies equally to regulations and to statutes. See Reno v. Flores, 507 U.S. 292, 300-301 (1993).

[8] See also 2 U. S. C. § 454 ("If any provision of this Act, or the application thereof to any person or circumstance, is held invalid, the validity of the remainder of the Act and the application of such provision to other persons and circumstances shall not be affected thereby" (emphasis added)); 5 U. S. C. § 806(b) (similar); 6 U. S. C. § 102 (2000 ed., Supp. II) (similar); 7 U. S. C. § 136x (similar); 15 U. S. C. § 79z-6 (similar); 29 U. S. C. § 114 (similar); 21 U. S. C. § 901 ("If a provision of this chapter is held invalid, all valid provisions that are severable shall remain in effect").

[9] See 2 U. S. C. § 454 ("If any provision of this Act, or the application thereof to any person or circumstance, is held invalid, the validity of the remainder of the Act and the application of such provision to other persons and circumstances shall not be affected thereby" (emphasis added)); 5 U. S. C. § 806(b) (similar); 6 U. S. C. § 102 (2000 ed., Supp. II) (similar); 7 U. S. C. § 136x (similar); 15 U. S. C. § 79z-6 (similar); 29 U. S. C. § 114 (similar); 21 U. S. C. § 901 (in relevant part, "[i]f a provision of this chapter is held invalid in one or more of its applications, the provision shall remain in effect in all its valid applications that are severable"); see also Vermeule 1950, n. 26 ("There is a common misconception that severability analysis refers only to the severance of provisions or subsections enumerated or labeled independently in the official text of the statute. In fact, however, severability problems arise not only with respect to different sections, clauses or provisions of a statute, but also with respect to applications of a particular statutory provision when some (but not all) of those applications are unconstitutional"); Stern, Separability and Separability Clauses in the Supreme Court, 51 Harv. L. Rev. 76, 78-79 (1937) ("One [type of severability question] relates to situations in which some applications of the same language in a statute are valid and other applications invalid").

[10] I do, however, agree with JUSTICE STEVENS that JUSTICE BREYER grossly distorts severability analysis by using severability principles to determine which provisions the Court should strike as unconstitutional. Ante, at 281-284 (STEVENS, J., dissenting in part). JUSTICE BREYER's severability analysis asks which provisions must be cut from the statute to fix the constitutional problem. Ante, at 245-249, 258 (opinion of the Court). Normally, however, a court (1) declares a provision or application unconstitutional, using substantive constitutional doctrine (not severability doctrine), and only then (2) asks (under severability principles) whether the remainder of the Act can be left standing. JUSTICE BREYER skips the first step, which is a necessary precursor to proper severability analysis.