The controlling law governing this case is crystal clear, but today the Court blithely discards any concept of stare decisis, trespasses gravely on the authority of the states, and distorts the concept of proportionality of punishment by tearing it from its moorings in capital cases. Only three Terms ago, we held in Rummel v. Estelle, 445 U. S. 263 (1980), that a life sentence imposed after only a third nonviolent felony conviction did not constitute cruel and unusual punishment under the Eighth Amendment. Today, the Court ignores its recent precedent and holds that a life sentence imposed after a seventh felony conviction constitutes cruel and unusual punishment under the Eighth Amendment. Moreover, I reject the fiction that all Helm’s crimes were innocuous or nonviolent. Among his felonies were three burglaries and a third conviction for drunken driving. By comparison Rummel was a relatively “model citizen.” Although today’s holding cannot rationally be reconciled with Rummel, the Court does not purport to overrule Rummel. I therefore dissent.
> — I
□>
The Court’s starting premise is that the Eighth Amendment’s Cruel and Unusual Punishments Clause “prohibits not *305only barbaric punishments, but also sentences that are disproportionate to the crime committed.” Ante, at 284. What the Court means is that a sentence is unconstitutional if it is more severe than five Justices think appropriate. In short, all sentences of imprisonment are subject to appellate scrutiny to ensure that they are “proportional” to the crime committed.
The Court then sets forth three assertedly “objective” factors to guide the determination of whether a given sentence of imprisonment is constitutionally excessive: (1) the “gravity of the offense and the harshness of the penalty,” ante, at 290-291; (2) a comparison of the sentence imposed with “sentences imposed on other criminals in the same jurisdiction,” ante, at 291 (emphasis added); (3) and a comparison of “the sentences imposed for commission of the same crime in other jurisdictions.” Ante, at 291-292 (emphasis added). In applying this analysis, the Court determines that respondent
“has received the penultimate sentence for relatively minor criminal conduct. He has been treated more harshly than other criminals in the State who have committed more serious crimes. He has been treated more harshly than he would have been in any other jurisdiction . . . .” Ante, at 303. (Emphasis added.)
Therefore, the Court concludes, respondent’s sentence is “significantly disproportionate to his crime, and is . . . prohibited by the Eighth Amendment.” This analysis is completely at odds with the reasoning of our recent holding in Rummel, in which, of course, Justice Powell dissented.
B
The facts in Rummel bear repeating. Rummel was convicted in 1964 of fraudulent use of a credit card; in 1969, he was convicted of passing a forged check; finally, in 1973 Rummel was charged with obtaining money by false pretenses, which is also a felony under Texas law. These three offenses were indeed nonviolent. Under Texas’ recidivist *306statute, which provides for a mandatory life sentence upon conviction for a third felony, the trial judge imposed a life sentence as he was obliged to do after the jury returned a verdict of guilty of felony theft.
Rummel, in this Court, advanced precisely the same arguments that respondent advances here; we rejected those arguments notwithstanding that his case was stronger than respondent’s. The test in Rummel which we rejected would have required us to determine on an abstract moral scale whether Rummel had received his “just deserts” for his crimes. We declined that invitation; today the Court accepts it. Will the Court now recall Rummel’s case so five Justices will not be parties to “disproportionate” criminal justice?
It is true, as we acknowledged in Rummel, that the “Court has on occasion stated that the Eighth Amendment prohibits imposition of a sentence that is grossly disproportionate to the severity of the crime.” 445 U. S., at 271. But even a cursory review of our cases shows that this type of proportionality review has been carried out only in a very limited category of cases, and never before in a case involving solely a sentence of imprisonment. In Rummel, we said that the proportionality concept of the capital punishment cases was inapposite because of the “unique nature of the death penalty . . . .” Id., at 272. “Because a sentence of death differs in kind from any sentence of imprisonment, no matter how long, our decisions applying the prohibition of cruel and unusual punishments to capital cases are of limited assistance in deciding the constitutionality of the punishment meted out to Rummel.” Ibid.
The Rummel Court also rejected the claim that Weems v. United States, 217 U. S. 349 (1910), required it to determine whether Rummel’s punishment was “disproportionate” to his crime. In Weems, the Court had struck down as cruel and unusual punishment a sentence of cadena temporal imposed by a Philippine Court. This bizarre penalty, which was un*307known to Anglo-Saxon law, entailed a minumum of 12 years’ imprisonment chained day and night at the wrists and ankles, hard and painful labor while so chained, and a number of “accessories” including lifetime civil disabilities. In Rummel the Court carefully noted that “[Weems’] finding of dispropor-tionality cannot be wrenched from the facts of that case.” 445 U. S., at 273.1
The lesson the Rummel Court drew from Weems and from the capital punishment cases was that the Eighth Amendment did not authorize courts to review sentences of imprisonment to determine whether they were “proportional” to the crime. In language quoted incompletely by the Court, ante, at 288-289, n. 14, the Rummel Court stated:
“Given the unique nature of the punishments considered in Weems and in the death penalty cases, one could argue without fear of contradiction by any decision of this Court that for crimes concededly classified and classifiable as felonies, that is, as punishable by significant terms of imprisonment in a state penitentiary, the length of the sentence actually imposed is purely a matter of legislative prerogative.” 445 U. S., at 274. (Emphasis added.)
Five Justices joined this clear and precise limiting language.
In context it is clear that the Rummel Court was not merely summarizing an argument, as the Court suggests, ante, at 288-289, n. 14, but was stating affirmatively the rule of law laid down. This passage from Rummel is followed by an explanation of why it is permissible for courts to review sentences of death or bizarre physically cruel punishments as in Weems, but not sentences of imprisonment. 445 U. S., at 274-275. The Rummel Court emphasized, as has every *308opinion in capital cases in the past decade, that it was possible to draw a “bright line” between “the punishment of death and the various other permutations and commutations of punishment short of that ultimate sanction”; similarly, a line could be drawn between the punishment in Weems and “more traditional forms of imprisonment imposed under the Anglo-Saxon system.” 445 U. S., at 275. However, the Rummel Court emphasized that drawing lines between different sentences of imprisonment would thrust the Court inevitably “into the basic line-drawing process that is pre-eminently the province of the legislature” and produce judgments that were no more than the visceral reactions of individual Justices. Ibid.
The Rummel Court categorically rejected the very analysis adopted by the Court today. Rummel had argued that various objective criteria existed by which the Court could determine whether his life sentence was proportional to his crimes. In rejecting Rummel’s contentions, the Court explained why each was insufficient to allow it to determine in an objective manner whether a given sentence of imprisonment is proportionate to the crime for which it is imposed.
First, it rejected the distinctions Rummel tried to draw between violent and nonviolent offenses, noting that “the absence of violence does not always affect the strength of society’s interest in deterring a particular crime or in punishing a particular criminal.” Ibid. Similarly, distinctions based on the amount of money stolen are purely “subjective” matters of line drawing. Id., at 275-276.
Second, the Court squarely rejected Rummel’s attempt to compare his sentence with the sentence he would have received in other States — an argument that the Court today accepts. The Rummel Court explained that such comparisons are flawed for several reasons. For one, the recidivist laws of the various states vary widely. “It is one thing for a court to compare those States that impose capital punishment for a *309specific offense with those States that do not. It is quite another thing for a court to attempt to evaluate the position of any particular recidivist scheme within Rummers complex matrix.” Id., at 280 (citation and footnote omitted). Another reason why comparison between the recidivist statutes of different states is inherently complex is that some states have comprehensive provisions for parole and others do not. Id., at 280-281. Perhaps most important, such comparisons trample on fundamental concepts of federalism. Different states surely may view particular crimes as more or less severe than other states. Stealing a horse in Texas may have different consequences and warrant different punishment than stealing a horse in Rhode Island or Washington, D. C. Thus, even if the punishment accorded Rummel in Texas were to exceed that which he would have received in any other state,
“that severity hardly would render Rummel’s punishment ‘grossly disproportionate’ to his offenses or to the punishment he would have received in the other States. . .. Absent a constitutionally imposed uniformity inimical to traditional notions of federalism, some State will always bear the distinction of treating particular offenders more severely than any other State.” Id., at 281-282. (Emphasis added.)
Finally, we flatly rejected Rummel’s suggestion that we measure his sentence against the sentences imposed by Texas for other crimes:
“Other crimes, of course, implicate other societal interests, making any such comparison inherently speculative. . . .
“. . . Once the death penalty and other punishments different in kind from fine or imprisonment have been put to one side, there remains little in the way of objective standards for judging whether or not a life sentence imposed under a recidivist statute for several separate *310felony convictions not involving ‘violence’ violates the cruel-and-unusual-punishment prohibition of the Eighth Amendment.” Id., at 282-283, n. 27.
Rather, we held that the severity of punishment to be accorded different crimes was peculiarly a matter of legislative policy. Ibid.
In short, Rummel held that the length of a sentence of imprisonment is a matter of legislative discretion; this is so particularly for recidivist statutes. I simply cannot understand how the Court can square Rummel with its holding that “a criminal sentence must be proportionate to the crime for which the defendant has been convicted.” Ante, at 290.2
If there were any doubts as to the meaning of Rummel, they were laid to rest last Term in Hutto v. Davis, 454 U. S. 370 (1982) (per curiam). There a United States District Court held that a 40-year sentence for the possession of nine ounces of marihuana violated the Eighth Amendment. The District Court applied almost exactly the same analysis adopted today by the Court. Specifically, the District Court stated:
“After examining the nature of the offense, the legislative purpose behind the punishment, the punishment in . . . Virginia [the sentencing jurisdiction] for other offenses, and the punishment actually imposed for the same or similar offenses in Virginia, this court must necessarily conclude that a sentence of forty years and twenty thousand dollars in fines is so grossly out of proportion to the severity of the crimes as to constitute cruel and unusual punishment in violation of the Eighth Amendment of the *311United States Constitution.” Davis v. Zahradnick, 432 F. Supp. 444, 453 (WD Va. 1977).
The Court of Appeals sitting en banc affirmed. Davis v. Davis, 646 F. 2d 123 (CA4 1981) (per curiam). We reversed in a brief per curiam opinion, holding that Rummel had disapproved each of the “objective” factors on which the District Court and en banc Court of Appeals purported to rely. 454 U. S., at 373. It was therefore clear error for the District Court to have been guided by these factors, which, paradoxically, the Court adopts today.
Contrary to the Court’s interpretation of Hutto, see ante, at 289-290, and n. 17, and 303-304, n. 32, the Hutto Court did not hold that the District Court miscalculated in finding Davis’ sentence disproportionate to his crime. It did not hold that the District Court improperly weighed the relevant factors. Rather, it held that the District Court clearly erred in even embarking on a determination whether the sentence was “disproportionate” to the crime. Hutto makes crystal clear that under Rummel it is error for appellate courts to second-guess legislatures as to whether a given sentence of imprisonment is excessive in relation to the crime,3 as the Court does today, ante, at 295-303.
I agree with what the Court stated only days ago, that “the doctrine of stare decisis, while perhaps never entirely persuasive on a constitutional question, is a doctrine that demands respect in a society governed by the rule of law.” City of Akron v. Akron Center for Reproductive Health, *312Inc., 462 U. S. 416, 419-420 (1983). While the doctrine of stare decisis does not absolutely bind the Court to its prior opinions, a decent regard for the orderly development of the law and the administration of justice requires that directly controlling cases be either followed or candidly overruled.4 Especially is this so with respect to two key holdings, neither more than three years old.
II
Although historians and scholars have disagreed about the Framers’ original intentions, the more common view seems to be that the Framers viewed the Cruel and Unusual Punishments Clause as prohibiting the kind of torture meted out during the reign of the Stuarts.5 Moreover, it is clear that *313until 1892, over 100 years after the ratification of the Bill of Rights, not a single Justice of this Court even asserted the doctrine adopted for the first time by the Court today. The prevailing view up to now has been that the Eighth Amendment reaches only the mode of punishment and not the length of a sentence of imprisonment.6 In light of this history, it is disingenuous for the Court blandly to assert that “[t]he constitutional principle of proportionality has been recognized explicitly in this Court for almost a century.” Ante, at 286. That statement seriously distorts history and our cases.
This Court has applied a proportionality test only in extraordinary cases, Weems being one example and the line of capital cases another. See, e. g., Coker v. Georgia, 433 U. S. 584 (1977); Enmund v. Florida, 458 U. S. 782 (1982). The Court’s reading of the Eighth Amendment as restricting legislatures’ authority to choose which crimes to punish by death rests on the finality of the death sentence. Such scrutiny is not required where a sentence of imprisonment is imposed after the State has identified a criminal offender whose record shows he will not conform to societal standards.
*314The Court’s traditional abstention from reviewing sentences of imprisonment to ensure that punishment is “proportionate” to the crime is well founded in history, in prudential considerations, and in traditions of comity. Today’s conclusion by five Justices that they are able to say that one offense has less “gravity” than another is nothing other than a bald substitution of individual subjective moral values for those of the legislature. Nor, as this case well illustrates, are we endowed with Solomonic wisdom that permits us to draw principled distinctions between sentences of different length for a chronic “repeater” who has demonstrated that he will not abide by the law.
The simple truth is that “[n]o neutral principle of adjudication permits a federal court to hold that in a given situation individual crimes are too trivial in relation to the punishment imposed.” Rummel v. Estelle, 568 F. 2d 1193, 1201-1202 (CA5) (Thornberry, J., dissenting), vacated, 587 F. 2d 651 (1978) (en banc), aff’d, 445 U. S. 263 (1980). The apportionment of punishment entails, in Justice Frankfurter’s words, “peculiarly questions of legislative policy.” Gore v. United States, 357 U. S. 386, 393 (1958). Legislatures are far better equipped than we are to balance the competing penal and public interests and to draw the essentially arbitrary lines between appropriate sentences for different crimes.
By asserting the power to review sentences of imprisonment for excessiveness the Court launches into uncharted and unchartable waters. Today it holds that a sentence of life imprisonment, without the possibility of parole, is excessive punishment for a seventh allegedly “nonviolent” felony. How about the eighth “nonviolent” felony? The ninth? The twelfth? Suppose one offense was a simple assault? Or selling liquor to a minor? Or statutory rape? Or price fixing? The permutations are endless and the Court’s opinion is bankrupt of realistic guiding principles. Instead, it casually lists several allegedly “objective” factors and arbitrarily asserts that they show respondent’s sentence to be “signifi*315cantly disproportionate” to his crimes. Ante, at 303. Must all these factors be present in order to hold a sentence excessive under the Eighth Amendment? How are they to be weighed against each other? Suppose several states punish severely a crime that the Court views as trivial or petty? I can see no limiting principle in the Court’s holding.
There is a real risk that this holding will flood the appellate courts with cases in which equally arbitrary lines must be drawn. It is no answer to say that appellate courts must review criminal convictions in any event; up to now, that review has been on the validity of the judgment, not the sentence. The vast majority of criminal cases are disposed of by pleas of guilty,7 and ordinarily there is no appellate review in such cases. To require appellate review of all sentences of imprisonment — as the Court’s opinion necessarily does — will “administer the coup de grace to the courts of appeals as we know them.” H. Friendly, Federal Jurisdiction: A General View 36 (1973). This is judicial usurpation with a vengeance; Congress has pondered for decades the concept of appellate review of sentences and has hesitated to act.
1 — 1 H — i
Even if I agreed that the Eighth Amendment prohibits imprisonment “disproportionate to the crime committed,” ante, at 284, I reject the notion that respondent’s sentence is disproportionate to his crimes for, if we are to have a system of laws, not men, Rummel is controlling.
The differences between this case and Rummel are insubstantial. First, Rummel committed three truly nonviolent felonies, while respondent, as noted at the outset, committed seven felonies, four of which cannot fairly be characterized as “nonviolent. ” At the very least, respondent’s burglaries and his third-offense drunken driving posed real risk of serious *316harm to others. It is sheer fortuity that the places respondent burglarized were unoccupied and that he killed no pedestrians while behind the wheel. What would have happened if a guard had been on duty during the burglaries is a matter of speculation, but the possibilities shatter the notion that respondent’s crimes were innocuous, inconsequential, minor, or “nonviolent.” Four of respondent’s crimes, I repeat, had harsh potentialities for violence. Respondent, far more than Rummel, has demonstrated his inability to bring his conduct into conformity with the minimum standards of civilized society. Clearly, this difference demolishes any semblance of logic in the Court’s conclusion that respondent’s sentence constitutes cruel and unusual punishment although Rummel’s did not.
The Court’s opinion necessarily reduces to the proposition that a sentence of life imprisonment with the possibility of commutation, but without possibility of parole, is so much more severe than a life sentence with the possibility of parole that one is excessive while the other is not. This distinction does not withstand scrutiny; a well-behaved “lifer” in respondent’s position is most unlikely to serve for life.
It is inaccurate to say, as the Court does, ante, at 301-302, that the Rummel holding relied on the fact that Texas had a relatively liberal parole policy. In context, it is clear that the Rummel Court’s discussion of parole merely illustrated the difficulty of comparing sentences between different jurisdictions. 445 U. S., at 280-281. However, accepting the Court’s characterization of Rummel as accurate, the Court today misses the point. Parole was relevant to an evaluation of Rummel’s life sentence because in the “real world,” he was unlikely to spend his entire life behind bars. Only a fraction of “lifers” are not released within a relatively few years. In Texas, the historical evidence showed that a prisoner serving a life sentence could become eligible for parole in as little as 12 years. In South Dakota, the historical evidence shows that since 1964, 22 life sentences have been commuted to *317terms of years, while requests for commutation of 25 life sentences were denied. And, of course, those requests for commutation may be renewed.
In short, there is a significant probability that respondent will experience what so many “lifers” experience. Even assuming that at the time of sentencing respondent was likely to spend more time in prison than Rummel,8 that marginal difference is surely supported by respondent’s greater demonstrated propensity for crime — and for more serious crime at that.
HH C
It is indeed a curious business for this Court to so far intrude into the administration of criminal justice to say that a state legislature is barred by the Constitution from identifying its habitual criminals and removing them from the streets. Surely seven felony convictions warrant the conclusion that respondent is incorrigible. It is even more curious that the Court should brush aside controlling precedents that are barely in the bound volumes of the United States Reports. The Court would do well to heed Justice Black’s comments about judges overruling the considered actions of legislatures under the guise of constitutional interpretation:
“Such unbounded authority in any group of politically appointed or elected judges would unquestionably be sufficient to classify our Nation as a government of men, not the government of laws of which we boast. With a ‘shock the conscience’ test of constitutionality, citizens *318must guess what is the law, guess what a majority of nine judges will believe fair and reasonable. Such a test wilfully throws away the certainty and security that lies in a written constitution, one that does not alter with a judge’s health, belief, or his politics.” Boddie v. Connecticut, 401 U. S. 371, 393 (1971) (dissenting).
Other authorities have shared this interpretation of Weems v. United States. E. g., Packer, Making the Punishment Fit the Crime, 77 Harv. L. Rev. 1071, 1075 (1964).
Although Rummel v. Estelle, 446 U. S., at 274, n. 11, conceded that “a proportionality principle [might] come into play... if a legislature made overtime parking a felony punishable by life imprisonment,” the majority has not suggested that respondent’s crimes are comparable to overtime parking. Respondent’s seven felonies are far more severe than Rummel’s three.
Both Rummel and Hutto v. Davis, leave open the possibility that in extraordinary cases — such as a life sentence for overtime parking — it might be permissible for a court to decide whether the sentence is grossly disproportionate to the crime. I agree that the Cruel and Unusual Punishments Clause might apply to those rare cases where reasonable men cannot differ as to the inappropriateness of a punishment. In all other cases, we should defer to the legislature’s line-drawing. However, the Court does not contend that this is such an extraordinary case that reasonable men could not differ about the appropriateness of this punishment.
I do not read the Court’s opinion as arguing that respondent’s sentence of life imprisonment without possibility of parole is so different from Rummel’s sentence of life imprisonment with the possibility of parole as to permit it to apply the proportionality review used in the death penalty cases, e. g., Coker v. Georgia, 433 U. S. 684 (1977), to the former although not the latter. Nor would such an argument be tenable. As was noted in Woodson v. North Carolina, 428 U. S. 280, 306 (1976) (opinion of Stewart, Powell, and Stevens, JJ.):
“[T]he penalty of death is qualitatively different from a sentence of imprisonment, however long. Death, in its finality, differs more from life imprisonment than a 100-year prison term differs from one of only a year or two. Because of that qualitative difference, there is a corresponding difference in the need for reliability in the determination that death is the appropriate punishment in a specific case.”
The greater need for reliability in death penalty cases cannot support a distinction between a sentence of life imprisonment with possibility of parole and a sentence of life imprisonment without possibility of parole, especially when an executive commutation is permitted as in South Dakota.
Compare, e. g., Granucci, “Nor Cruel and Unusual Punishments Inflicted:” The Original Meaning, 57 Calif. L. Rev. 839 (1969); Schwartz, Eighth Amendment Proportionality Analysis and the Compelling Case of William Rummel, 71 J. Crim. L. & Criminology 378, 379-382 (1980); Katkin, Habitual Offender Laws: A Reconsideration, 21 Buffalo L. Rev. 99, 115 (1971), with, e. g., Wheeler, Toward a Theory of Limited Punishment: An Examination of the Eighth Amendment, 24 Stan. L. Rev. 838, *313853-855 (1972); Comment, The Eighth Amendment, Beccaria, and the Enlightenment: An Historical Justification for the Weems v. United States Excessive Punishment Doctrine, 24 Buffalo L. Rev. 783 (1975).
In 1892, the dissent in O’Neil v. Vermont, 144 U. S. 323, 339-340 (1892) (Field, J., dissenting), argued that the Eighth Amendment “is directed ... against all punishments which by their excessive length or severity are greatly disproportioned to the offenses charged.” Before and after O’Neil, most authorities thought that the Eighth Amendment reached only the mode of punishment and not the length of sentences. See, e. g., Note, 24 Harv. L. Rev. 54, 55 (1910). Even after Weems v. United States, 217 U. S. 349, was decided in 1910, it was thought unlikely that the Court would extend proportionality analysis to cases involving solely sentences of imprisonment. See Packer, supra n. 1, at 1075. Until today, not a single case of this Court applied the “excessive punishment” doctrine of Weems to a punishment consisting solely of a sentence of imprisonment, despite numerous opportunities to do so. E. g., Hutto v. Davis, 454 U. S. 370 (1982); Rummel v. Estelle, 445 U. S. 263 (1980); Badders v. United States, 240 U. S. 391 (1916); Graham v. West Virginia, 224 U. S. 616 (1912).
In 1972, nearly 90% of the convictions in federal courts followed pleas of guilty or nolo contedere. H. Friendly, Federal Jurisdiction: A General View 36 (1973).
No one will ever know if or when Rummel would have been released on parole since he was released in connection with a separate federal habeas corpus proceeding in 1980. On October 3, 1980, a Federal District Court granted Rummel’s petition for a writ of habeas corpus on the grounds of ineffective assistance of counsel. Rummel v. Estelle, 498 F. Supp. 793 (WD Tex. 1980). Rummel then pleaded guilty to theft by false pretenses and was sentenced to time served under the terms of a plea-bargaining agreement. Two-Bit Lifer Finally Freed — After Pleading Guilty, Chicago Tribune, Nov. 15, 1980, p. 2, col. 3.