with whom Justice White and Justice Blackmun join, dissenting.
In Gomez v. United States, 490 U. S. 858 (1989), this Court held that the Federal Magistrates Act does not authorize magistrates to conduct jury selection at a felony trial. In an *941amazing display of interpretive gymnastics, the majority twists, bends, and contorts the logic of Gomez, attempting to demonstrate that the consideration critical to our holding in that case was the defendant’s refusal to consent to magistrate jury selection. I find Gomez to be considerably less flexible. Our reasoning in Gomez makes clear that the absence or presence of consent is entirely irrelevant to the Federal Magistrates Act’s prohibition upon magistrate jury selection in a felony trial.
The majority’s reconstruction of Gomez is not only unsound, but also unwise. By discarding Gomez’s categorical prohibition of magistrate felony jury selection, the majority unnecessarily raises the troubling question whether this practice is consistent with Article III of the Constitution. To compound its error, the majority resolves the constitutional question in a manner entirely inconsistent with our controlling precedents. I dissent.
I
A
The majority purports to locate the source of a magistrate’s authority to conduct consented-to felony jury selection in the Act’s “additional duties” clause, which states that “[a] magistrate may be assigned such additional duties as are not inconsistent with the Constitution and laws of the United States.” 28 U. S. C. § 636(b)(3). Whether the additional duties clause authorizes a magistrate to conduct jury selection in a felony trial is a conventional issue of statutory interpretation. In Gomez, we held that “[t]he absence of a specific reference to jury selection in the statute, or indeed, in the legislative history, persuades us that Congress did not intend the additional duties clause to embrace this function.” 490 U. S., at 875-876 (footnote omitted). In my view, the existence of a defendant’s consent has absolutely no effect on that conclusion.
*942In Gomez, we rejected a literal reading of the additional duties clause that would have authorized magistrates to exercise any power not expressly prohibited by federal statute or the Constitution. See id., at 864-865. Relying on precedent and legislative history, we emphasized that the additional duties clause is to be read according to Congress’ intention that magistrates “handle subsidiary matters[,] [thereby] enabling] district judges to concentrate on trying cases.” Id., at 872.
“If district judges are willing to experiment with the assignment to magistrates of other functions in aid of the business of the courts, there will be increased time available to judges for the careful and unhurried performance of their vital and traditional adjudicatory duties, and a consequent benefit to both efficiency and the quality of justice in the Federal courts.” H. R. Rep. No. 94-1609, p. 12 (1976) (emphasis added) (1976 amendments to Federal Magistrates Act); accord, S. Rep. No. 371, 90th Cong., 1st Sess., 26 (1967) (Federal Magistrates Act of 1968).1
We identified two reasons in Gomez for inferring that Congress intended jury selection in felony trials to be one of the “vital and traditional adjudicatory duties” retained by district *943judges rather than delegated to magistrates. First, we noted that Congress felt it necessary to define expressly a magistrate’s limited authority to conduct misdemeanor and civil trials. See 28 U. S. C. §§ 636(a)(3), 636(c). We concluded that “th[is] carefully defined grant of authority to conduct trials of civil matters and of minor criminal cases” constituted “an implicit withholding of the authority to preside at a felony trial.” Gomez, 490 U. S., at 872. And in light of the traditional judicial and legislative understanding that jury selection is an essential component of a felony trial,2 we determined that Congress’ intention to deny magistrates the authority to preside at felony trials also extends to jury selection. See id., at 871-872.
In my view, this structural inference is not at all affected by a defendant’s consent. Under the Act, consent of the parties is a necessary condition of a magistrate’s statutory authority to preside at a civil or misdemeanor trial. See 18 U. S. C. § 3401(b); 28 U. S. C. § 636(c)(1). To hold, as the majority does, that a magistrate may likewise conduct jury selection in a felony trial so long as the defendant consents is to treat the magistrate’s authority in this part of the felony trial as perfectly coextensive with his authority in civil and misdemeanor trials — the reading of the Act that Gomez categorically Rejected.
*944The second basis for our conclusion in Gomez that Congress intended felony jury selection to be nondelegable was Congress’ failure expressly to provide for judicial review of magistrate jury selection in felony cases. The Federal Magistrates Act provides two separate standards of judicial review: “clearly erroneous or contrary to law” for magistrate resolution of nondispositive matters, see 28 U. S. C. § 636(b) (1)(A), and “de novo” for magistrate resolution of dispositive matters, see § 636(b)(1)(B) — (C). We deemed Congress’ failure to identify any standard of judicial review for jury selection in felony trials to be persuasive evidence of Congress’ intent that magistrates not perform this function. Gomez, supra, at 873-874.
Again, I fail to see how a defendant’s consent to a magistrate’s exercise of such authority can alter this inference. Congress said no more about the standard of review for consented-to magistrate jury selection than it did about the standard for unconsented-to magistrate jury selection. Nor does the majority identify anything in the statute to indicate the appropriate standard for consented-to magistrate jury selection.
The majority opines that “nothing in the statute precludes” judicial review, ante, at 939. However, it fails to explain how such review may be achieved. The majority’s silence is regrettable. In Gomez, we recognized that jury selection is most similar to the functions identified as “dispositive matters,” for which the Act prescribes a de novo review standard. 490 U. S., at 873. We expressed “serious doubts,” however, as to whether any review could be meaningfully conducted. Id., at 874.3 We likewise concluded that re*945examination of individual jurors by the district judge would not be feasible because “as a practical matter a second interrogation might place jurors on the defensive, engendering prejudices irrelevant to the facts adduced at trial.” Id., at 875, n. 29. These difficulties in providing effective review of magistrate jury selection were central to our construction of the Act in Gomez, yet they are essentially ignored today.4
In Gomez, we found confirmation of the inferences that we drew from the statutory text in “[t]he absence of a specific reference to jury selection in . . . the legislative history.” Id., at 875. See ante, at 930. The legislative history of the Act offers no more support for consented-to magistrate felony jury selection.5
In response to the paucity of support for its construction, the majority notes that in Gomez we “call[ed] attention” to a House Committee Report that “referred” to a letter from a district judge mentioning jury selection as a duty assigned to *946magistrates. Ante, at 935, n. 11. While the majority observes that the letter “ ‘suggested] that a magistrate selected juries only with consent of the parties,’” ibid., quoting Gomez, 490 U. S., at 875, n. 30 (emphasis added by majority), it neglects to record other salient facts that we noted about this letter. In particular, the letter was the “lone reference” in the entire legislative history to such authority. Ibid. (emphasis added). Moreover, the letter suggested that magistrate jury selection took place “perhaps only in civil trials” Id., at 876, n. 30 (emphasis added). Finally, as we pointed out in Gomez,
“[the letter] displays little concern about the validity of such assignments: 'How can we do all of this? We just do it. It’s not necessary that we find authority in black and white before we give something to the magistrate. . . . Sure we might get shot down once in a while by an appellate court. So what?’” Ibid, (citation omitted).
B
It is clear that the considerations that motivated our holding in Gomez compel the conclusion that the Federal Magistrates Act does not permit magistrate felony jury selection even when the defendant consents. I find the majority’s arguments to the contrary wholly unpersuasive.
According to the majority, “[t]his case differs critically from Gomez” because petitioner’s counsel consented to the delegation of jury selection to the Magistrate. Ante, at 932. Although it asserts that this factor was essential to our analysis, the majority fails to explain how consent has any bearing on the statutory power of a magistrate to conduct felony jury selection. As I have already indicated, the reasoning behind our conclusion in Gomez that Congress did not endow magistrates with jurisdiction to preside over felony jury selection had nothing to do with the defendant’s refusal to consent to such jurisdiction.
*947Unable to support its revisionist construction of the Act with what we said in Gomez, the majority seeks to bolster its construction by noting that, provided the parties consent, magistrates may conduct civil and misdemeanor trials and that “[t]hese duties are comparable in responsibility and importance to presiding over voir dire at a felony trial.” Ante, at 933. The majority’s analogy misses the point. The fact that Congress imposed the condition of consent on magistrates’ exercise of expressly-provided authority does not prove that Congress also authorized magistrates to conduct trial duties not expressly enumerated in the Federal Magistrates Act — such as supervision of felony jury selection. At most, these specifically enumerated grants of trial authority suggest that if Congress had intended to confer on magistrates authority to conduct felony jury selection, it xoould have predicated that authority on the parties’ consent. However, as I have already discussed, see supra, at 942-943, construing the Act as authorizing magistrates to conduct consented-to jury selection in felony cases merely because the Act authorizes consented-to jurisdiction in civil and misdemeanor cases is to draw an inference from Congress’ silence precisely opposite to the inference we drew in Gomez.6
*948Finally, the majority defends its construction of the additional duties clause by stating that it will permit “ ‘continue[d] innovative experimentations’ in the use of magistrates to improve the efficient administration” of the district courts. Ante, at 934. Taken literally, such a rationale admits of no limits, and for this reason it cannot function as a legitimate basis for construing the scope of a magistrate’s permissible “additional duties.” As in Gomez, we must give content to the additional duties clause by looking to Congress’ intention that magistrates be delegated administrative and other quasi-judicial tasks in order to free Article III judges to conduct trials, most particularly felony trials. See swpra, at 942. By creating authority for magistrates to preside over a “critical stage” of the felony trial, see Gomez, supra, at 873, merely because a defendant fails to request a judge, the majority completely misapprehends both Congress’ conception of the appropriate role to be played by magistrates and our analysis in Gomez.
II
I have outlined why I believe the only defensible construction of the Federal Magistrates Act is that jury selection in a felony trial can never be one of a magistrate’s “additional duties” — regardless of whether a defendant consents. But even if I believed that mine was only one of two “reasonable” interpretations, I would still reject the majority’s construction of the Act, because it needlessly raises a serious constitutional question: whether jury selection by a magistrate— *949even when a defendant consents — is consistent with Article III.
It is well established that we should “avoid an interpretation of a federal statute that engenders constitutional issues if a reasonable alternative interpretation poses no constitutional question.” Gomez, 490 U. S., at 864; accord, e. g., Edward J. DeBartolo Corp. v. Florida Gulf Coast Building & Construction Trades Council, 485 U. S. 568, 575 (1988); Commodity Futures Trading Comm’n v. Schor, 478 U. S. 833, 841 (1986); Ashwander v. TVA, 297 U. S. 288, 348 (1936) (Brandeis, J., concurring). Given the inherent complexity of Article III questions, the canon of constitutional avoidance should apply with particular force when an Article III issue is at stake. Cf. Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U. S. 50, 90 (1982) (Rehnquist, J., concurring in judgment) (“Particularly in an area of constitutional law such as that of ‘Art. Ill Courts,’ with its frequently arcane distinctions and confusing precedents, rigorous adherence to the principle that this Court should decide no more of a constitutional question than is absolutely necessary accords with both our decided cases and with sound judicial policy”).
Although this principle guided our analysis in Gomez, see 490 U. S., at 864, it is all but forgotten today. The majority simply dismisses altogether the seriousness of the underlying constitutional question: “[W]e have no trouble concluding that there is no Article III problem when a district court judge permits a magistrate to conduct voir dire in accordance with the defendant’s consent.” Ante, at 932. The majority’s self-confidence is unfounded. It is only by unacceptably manipulating our Article III teachings that the majority succeeds in avoiding the difficulty that attends its construction of the Act.
As the Court explained in Schor, Article Ill’s protections have two distinct dimensions. First, Article III “safeguard[s] litigants’ ‘right to have claims decided before judges who are free from potential domination by other branches of *950government.’” Schor, supra, at 848, quoting United States v. Will, 449 U. S. 200, 218 (1980). Second, Article III “serves as ‘an inseparable element of the constitutional system of checks and balances’ ” by preserving “the role of the Judicial Branch in our tripartite system” of government. Schor, supra, at 850, quoting Northern Pipeline, supra, at 58. Although parties may waive their personal guarantee of an independent Article III adjudicator, Schor, supra, at 848, parties may not waive Article III’s structural guarantee.
“Article III, §1, safeguards the role of the Judicial Branch in our tripartite system by barring congressional attempts ‘to transfer jurisdiction [to non-Article III tribunals] for the purpose of emasculating’ constitutional courts .... To the extent that this structural principle is implicated in a given case, the parties cannot by consent cure the constitutional difficulty for the same reason that the parties by consent cannot confer on federal courts subject-matter jurisdiction beyond the limitations imposed by Article III, §2. When these Article III limitations are at issue, notions of consent and waiver cannot be dispositive because the limitations serve institutional interests that the parties cannot be expected to protect.” 478 U. S., at 850-851 (emphasis added; citations omitted).
In Gomez, we recognized and attempted to accommodate “abiding concerns regarding the constitutionality of delegating felony trial duties to magistrates.” See 490 U. S., at 863. Because jury selection is “a critical stage” of the felony trial, see id., at 873, there is a serious question, as several Courts of Appeals have noted, whether allowing a magistrate to conduct felony jury selection “impermissibly intrude[s] on the province of the judiciary,” Schor, supra, at 851-852. See United States v. Trice, 864 F. 2d 1421, 1426 (CA8 1988), cert. dism’d, 491 U. S. 914 (1989); United States v. Ford, 824 F. 2d 1430, 1434-1435 (CA5 1987) (en banc), cert. denied, 484 U. S. 1034 (1988).
*951Indeed, this problem admits of no easy solution. This Court’s decision in United States v. Raddatz, 447 U. S. 667 (1980), suggests that delegation of Article III powers to a magistrate is permissible only if the ultimate determinations on the merits of delegated matters are made by the district judge. See id., at 683 (“[Although the [Federal Magistrates Act] permits the district court to give to the magistrate’s proposed findings of fact and recommendations ‘such weight as [their] merit commands and the sound discretion of the judge warrants,’ that delegation does not violate Art. Ill so long as the ultimate decision is made by the district court” (emphasis added; citation omitted)).7 In Schor, we likewise emphasized the availability of de novo judicial review in upholding the performance of core Article III powers by an Article I tribunal. See 478 U. S., at 853. But this means of satisfying the Constitution is not available here. For, as I have noted, swpra, at 944, the Federal Magistrates Act does not expressly provide for judicial review of felony jury selection, and in Gomez we expressed “serious doubts” whether such review was even possible. See 490 U. S., at 874.
The majority contends that magistrate jury selection raises no Article III structural difficulties, because “‘the entire process takes place under the district court’s total control and jurisdiction.’” Ante, at 937, quoting Raddatz, supra, at 681. However, as Raddatz and Schor underscore, the requirement of “the district court’s total control and jurisdiction” must include the availability of meaningful judicial review of the magistrate’s actual rulings at jury selection. The majority’s observation that “nothing in the statute precludes a dis*952trict court from providing the review that the Constitution requires,” ante, at 939, is equally unavailing. The critical question for Article III purposes is whether meaningful judicial review of magistrate felony jury selection can be accomplished. The majority does not answer this question, and Gomez strongly suggests that it cannot.
Because it ignores the teachings of Raddatz and Schor, the majority’s analysis of the Article III difficulty posed by its construction of the Federal Magistrates Act raises the question whether these decisions remain good law. This consequence is particularly unfortunate, because, as I have set forth above, the most coherent reading of the Federal Magistrates Act avoids these problems entirely.
I dissent.
This theme pervades the Act’s legislative history. See, e. g., S. Rep. No. 96-74, p. 3 (1979) (1979 amendments to Federal Magistrates Act) (“In enacting the Federal Magistrates Act in 1968, the Congress clearly intended that the magistrate should be a judicial officer whose purpose was to assist the district judge to the end that the judge could have more time to preside at the trial of cases”); H. R. Rep. No. 94-1609, p. 6 (1976) (same); S. Rep. 94-625, p. 6 (1976) (1976 amendments to Federal Magistrates Act) (“Without the assistance furnished by magistrates . . . the judges of the district courts would have to devote a substantial portion of their available time to various procedural steps rather than to the trial itself”); see also S. Rep. No. 371, 90th Cong., 1st Sess., 9 (1967) (Federal Magistrates Act is intended “to cull from the ever-growing workload of the U. S. district courts matters that are more desirably performed by a lower tier of judicial officers”).
As we have observed, “‘“[W]here the indictment is for a felony, the trial commences at least from the time when the work of empanelling the jury begins.’”” Gomez v. United States, 490 U. S. 858, 873 (1989), quoting Lewis v. United States, 146 U. S. 370, 374 (1892), quoting Hopt v. Utah, 110 U. S. 574, 578 (1884). Moreover, “[j]ury selection is the primary means by which a court may enforce a defendant’s right to be tried by a jury free from ethnic, racial, or political prejudice, or predisposition about the defendant’s culpability.” Gomez, supra, at 873 (citations omitted). We discerned Congress’ recognition of this understanding from its passage of the Speedy Trial Act, 18 U. S. C. § 3161, and from its placement of rules relating to juries and jury selection in a chapter of the Federal Rules of Criminal Procedure entitled “Trial.” See Gomez, supra, at 873, citing Fed. Rules Crim. Proc. 23 and 24.
“To detect prejudices, the examiner — often, in the federal system, the court — must elicit from prospective jurors candid answers about intimate details of their lives. The court further must scrutinize not only spoken words but also gestures and attitudes of all participants to ensure the jury’s impartiality. But only words can be preserved for review; no tran*945script can recapture the atmosphere of the voir dire, which may persist throughout the trial.” Gomez, supra, at 874-875 (citations omitted).
The majority concedes that magistrate jury selection “may be difficult for a judge to review with infallible accuracy.” Ante, at 935, n. 12. But it dismisses any concerns with respect to the difficulty of effective judicial review, stating that the defendant can eliminate the need for judicial review altogether by simply declining to consent to magistrate jury selection. Ante, at 935, and n. 12. This rationalization misses the point. Insofar as the Federal Magistrates Act insists that magistrate functions be subject to judicial review, the impossibility of effective review is reason not to construe the additional duties clause as authorizing magistrates to conduct felony jury selection, regardless of whether the parties consent. See Gomez, supra, at 874-875.
In Gomez, we noted that Committee Reports accompanying the 1976 and 1979 amendments to the Magistrates Act contained charts cataloging-magistrate functions. In determining Congress’ understanding of the permissible scope of magistrate duties, we found it relevant that not one of the charts mentioned jury selection. See Gomez, 490 U. S., at 875, n. 30 (citing H. R. Rep. No. 96-287, pp. 4-5 (1979); S. Rep. No. 96-74, at 3; H. R. Rep. No. 94-1609, at 7; S. Rep. No. 94-625, at 5). Needless to say, the charts also contain no mention of jury selection where the parties have consented to magistrate supervision.
Even if I were to accept the majority’s conclusion that the scope of a magistrate’s authority under the additional duties clause turns on litigant consent, I still could not accept the majority’s assumption that there was effective consent in this case. Because the additional duties clause contains no language predicating delegation of an additional duty upon litigant consent, it likewise contains nothing indicating what constitutes “consent” to the delegation of an additional duty. I would think, however, that the standard governing a party’s consent to delegation of a portion of a felony trial under the additional duties clause should be at least as strict as that governing delegation of a misdemeanor trial to a magistrate. Under the Act, before a magistrate can conduct a misdemeanor trial, the magistrate must explain to the defendant that he has a right to a trial before a district court judge. If the defendant elects to proceed before the magistrate, the defendant must consent in writing. See 18 U. S. C. § 3401(b); see also 28 U. S. C. § 636(a)(3) (incorporating requirements of 18 U. S. C. § 3401 into the Federal Magistrates Act). The procedural safeguard of written con*948sent by the defendant “ ‘show[s] a statutory intent to preserve trial before the district judge as the principal — rather than an elective or alternative— mode of proceeding in minor offense cases.’ ” Gomez, supra, at 872, n. 24, (quoting 114 Cong. Rec. 27342 (1968) (remarks of Rep. Poff)). In this case, the defendant did not consent in writing; in fact, the defendant did not proffer consent in any form. Instead, what the majority accepts as sufficient consent were merely verbal remarks made by defense counsel at a pretrial conference and jury selection. See App. 2, 5.
The majority seeks to evade this difficulty by pronouncing that Justice Blackmun’s concurring opinion in Raddatz now “eontrol[s]” the constitutional analysis of a delegation of Article III duties to a magistrate. Ante, at 938. Justice Blackmun’s opinion in Raddatz, however, offers little repose for the majority, for Justice Blackmun likewise identifies the availability of judicial review as a necessary predicate of the constitutionality of any delegation of Article III duties to a magistrate. See United States v. Raddatz, 447 U. S., at 685 (Blackmun, J., concurring).