dissenting.
Title 8 U. S. C. § 1231(a)(6) states that aliens whom the Secretary of Homeland Security has ordered removed “may be detained beyond the removal period.” Nevertheless, in Zadvydas v. Davis, 533 U. S. 678 (2001), this Court construed this provision “to contain an implicit ‘reasonable time’ limitation” on the Secretary’s power to detain admitted aliens “[b]ased on our conclusion that indefinite detention of” those aliens “would raise serious constitutional concerns.” Id., at 682. “Aliens who have not yet gained initial admission to this country,” the Court assured us, “would present a very different question.” Ibid.
Today, the Court holds that this constitutional distinction — which “made all the difference” to the Zadvydas Court, id., at 693 — is actually irrelevant, because “[t]he operative language of § 1231(a)(6). .. applies without differentiation to all three categories of aliens that are its subject.” Ante, at 378. While I wholeheartedly agree with the Court’s fidelity to the text of § 1231(a)(6), the Court’s analysis cannot be squared with Zadvydas. And even if it could be so squared, Zadvydas was wrongly decided and should be overruled. I respectfully dissent.
I
I begin by addressing the majority’s interpretation of Zadvydas. The Court’s interpretation is not a fair reading of that case. It is also not required by any sound principle of statutory construction of which I am aware. To the con*389trary, what drives the majority’s reading is a novel “lowest common denominator” principle. Ante, at 380.
A
The majority’s reading of Zadvydas is implausible. Zadvydas held that interpreting § 1231(a)(6) to authorize indefinite detention of admitted aliens later found removable would raise serious due process concerns. 533 U. S., at 690-696. The Court therefore read the statute to permit the Attorney General (now the Secretary of Homeland Security) to detain admitted aliens only as long as reasonably necessary to remove them from the country. Id., at 699.
The majority concedes that Zadvydas explicitly reserved the question whether its statutory holding as to admitted aliens applied equally to inadmissible aliens. Ante, at 379. This reservation was front and center in Zadvydas. It appeared in the introduction and is worth repeating in full:
“In these cases, we must decide whether [§ 1231(a)(6)] authorizes the Attorney General to detain a removable alien indefinitely beyond the removal period or only for a period reasonably necessary to secure the alien’s removal. We deal here with aliens, who were admitted to the United States but subsequently ordered removed. Aliens who have not yet gained initial admission to this country would present a very different question. Based on our conclusion that indefinite detention of aliens in the former category would raise serious constitutional concerns, we construe the statute to contain an implicit ‘reasonable time’ limitation, the application of which is subject to federal-court review.” 533 U. S., at 682 (citation omitted; emphasis in original).
The Court reserved this question because the constitutional questions raised by detaining inadmissible aliens are different from those raised by detaining admitted aliens. It stated that the detention period in § 1231(a)(6) was limited *390because it “read [the statute] in light of the Constitution’s demands.” Id., at 689. And it repeatedly emphasized constitutional distinctions among various groups of aliens, for which § 1231(a)(6) makes no distinctions. See id., at 693-694 (noting the different constitutional, considerations applicable to inadmissible and admissible aliens); id., at 695 (noting that “the cases before us [do not] require us to consider the political branches’ authority to control entry into the United States”); id., at 696 (noting that the opinion did not “consider terrorism or other special circumstances where special arguments might be made for forms of preventive detention and for heightened deference to the judgments of the political branches with respect to matters of national security”).
The majority’s reading of Zadvydas is inconsistent with these qualifications. If it were true that Zadvydas’ interpretation of § 1231(a)(6) applied to all aliens regardless of the constitutional concerns involved in each case, then the question of how § 1231(a)(6) applies to them would not be “very different” depending on the alien before the Court. The question would be trivial because the text of § 1231(a)(6) plainly does not distinguish between admitted and nonad-mitted aliens. There would also have been no need for the Court to go out of its way to leave aside “terrorism or other special circumstances,” ibid., or to disavow “consider[ation of] the political branches’ authority to control entry into the United States,” id., at 695, for the construction the majority extracts from Zadvydas would have applied across the board, ibid. And the Court’s rationalization that its construction would therefore “leave no unprotected spot in the Nation’s armor,” id., at 695-696 (internal quotation marks omitted), would have been incorrect. The constitutional distinctions that pervade Zadvydas are evidence that the “very different” statutory question it reserved turned on them.
The Zadvydas Court thus tethered its reading of § 1231(a)(6) to the specific class of aliens before it. The term this Court read into the statute was not simply a presump*391tive 6-month period, but a presumptive 6-month period for admitted aliens. Its reading of the statute “in light of the Constitution’s demands,” id., at 689, that is, depended on the constitutional considerations at work in “the cases before [it],” id., at 695 (emphasis added). One would expect the Court today, then, to follow the same two-step procedure it employed in Zadvydas. It should first ask whether the statute is ambiguous and, if so, whether one of the possible interpretations raises constitutional doubts as applied to respondent Martinez and petitioner Benitez. Step one is dictated by Zadvydas: Section 1231(a)(6) is not clear on whether it permits indefinite detention. The Court should then move to the second step and ask whether either of the statute’s possible interpretations raises constitutional doubts as applied to Benitez and Martinez. If so, the Court would apply avoidance to adopt the interpretation free from constitutional doubt (as Zadvydas itself did).
The Court’s reasons for departing from this reading of Zadvydas are unpersuasive. The Court says that its reading is necessary to avoid “inventing] a statute rather than interpreting] one,” ante, at 378; to preclude “giving the same detention provision a different meaning” depending on the aliens before the Court, ante, at 380 (emphasis in original); and to forestall establishing “the dangerous principle that judges can give the same statutory text different meanings in different cases,” ante, at 386. I agree that we should adopt none of these principles, but this is no warrant for the reading of Zadvydas that the majority advocates. Zadvydas established a single and unchanging, if implausible, meaning of § 1231(a)(6): that the detention period authorized by § 1231(a)(6) depends not only on the circumstances surrounding a removal, but also on the type of alien ordered removed.
I grant that this understanding of Zadvydas could result in different detention periods for different classes of aliens— indefinite detention for some, limited detention for others. *392But it does not follow that this reads the meaning of the statute to “change” depending on the alien involved, any more than the meaning of the statute could be said to “change” simply because the time that is “reasonably necessary to effect removal” may differ depending on the type of alien involved, as both the Court’s opinion, ante, at 386, and Justice O’Connor’s concurring opinion, ante, at 387, concede it may. A statute’s sense is the same even if what it requires depends on factual context.
In support of its reading of Zadvydas, the Court relies on a statement in a dissent in Zadvydas that § 1231(a)(6) could not be given a different reading for inadmissible aliens. Ante, at 379-380 (citing 533 U. S., at 710-711, 717 (opinion of Kennedy, J.)). That dissenting view, as the very quotation the majority stresses demonstrates, rested on the dissent’s premise that “it is not a plausible construction of § 1231(a)(6) to imply a time limit as to one class but not to another.” Id., at 710. But the Zadvydas majority disagreed with that assumption and adopted a contrary interpretation of § 1231(a)(6). For as -the dissent recognized, Zadvydas’ “logic might be that inadmissible and removable aliens can be treated differently.” Ibid. That was Zadvydas’ logic precisely, as its repeated statements limiting its decision to inadmissible aliens show. To interpret Zadvydas properly, we must take its logic as given, not the logic of the reductio ad absurdum of Zadvydas that I joined in dissent.
B
The majority strains to recharacterize Zadvydas because it thinks that “[i]t is not at all unusual to give a statute’s ambiguous language a limiting construction called for by one of the statute’s applications, even though other of the statute’s applications, standing alone, would not support the same limitation.” Ante, at 380. In other words, it claims, “[t]he lowest common denominator, as it were, must govern.” Ibid. I disagree.
*393As an initial matter, this principle is inconsistent with Zadvydas itself. As explained above, the limiting construction Zadvydas adopted as to admitted aliens does not necessarily govern the other applications of § 1231(a)(6). If the majority is correct that the “lowest common denominator” governs, then the careful distinction Zadvydas drew between admitted aliens and nonadmitted aliens was irrelevant at best and misleading at worst. Under this reading, Zadvydas would have come out the same way even if it had involved inadmissible aliens, for the “lowest common denominator” of the statute remains the same regardless of the identity of the alien before the Court. Again, this understanding of Zadvydas is implausible.
Beyond Zadvydas, the Court offers scant support for the idea that statutes should be stripped down to their “lowest common denominator[s].” It attempts to distinguish Jinks v. Richland County, 538 U. S. 456 (2003), and Raygor v. Regents of Univ. of Minn., 534 U. S. 533 (2002), ante, at 383, and n. 6, yet these cases employed exactly the procedure that the majority today says is impermissible. They construed 28 U. S. C. § 1367(d),1 a tolling provision, to apply to States and political subdivisions of States only to the extent that doing so would raise a constitutional doubt as applied to either entity. Jinks was explicit on this point:
“Although we held in [Raygor] that § 1367(d) does not apply to claims filed in federal court against States but subsequently dismissed on sovereign immunity grounds, we did so to avoid interpreting the statute in a manner that would raise ‘serious constitutional doubt’ in light of our decisions protecting a State’s sovereign immunity *394from congressional abrogation .... [N]o such constitutional doubt arises from holding that petitioner’s claim against respondent — which is not a State, but a political subdivision of a State — falls under the definition of ‘any claim asserted under subsection (a) [of § 1367].’ ” 538 U. S., at 466 (emphasis in original).
This passage reads the meaning of § 1367(d) — which applies to “any claim asserted under subsection (a)” of §1367 — to hinge on the constitutional context. The Court is correct that Jinks and Raygor “hold that the single and unchanging disposition of § 1367(d) . . . does not apply to claims against States.” Ante, at 383. But as the Court concedes, Jinks reached that holding only after analyzing whether the constitutional doubts at issue in Raygor applied to the county defendant. Ante, at 383-384, n. 6. The Court’s failure to do the same here cannot be reconciled with Jinks and Raygor: the Court should ask whether the constitutional concerns that justified the requirement of a clear statement in Zadvydas apply as well to inadmissible aliens.
The Court’s “lowest common denominator” principle is also in tension with Salinas v. United States, 522 U. S. 52 (1997). There, we rejected an argument that the federal bribery statute, 18 U. S. C. § 666(a)(1)(B), should be construed to avoid constitutional doubts, in part on the ground that there was “no serious doubt about the constitutionality of § 666(a)(1)(B) as applied to the facts of this case.” 522 U. S., at 60 (emphasis added). Unlike the Court’s approach to avoidance today, we disclaimed examination of the constitutionality of applications not before the Court: “Whatever might be said about § 666(a)(l)(B)’s application in other cases, the application of § 666(a)(1)(B) . . . did not extend federal power beyond its proper bounds.” Id., at 61. The Court is mistaken that this passage in Salinas was a rejection of a constitutional argument on its merits. Ante, at 381, n. 5. Salinas, the petitioner, phrased his question presented solely in terms of the proper statutory interpretation of *395§ 666(a)(1)(B), Brief for Petitioner, O. T. 1996, No. 96-738, p. i, and never claimed that the statute was unconstitutional, see generally ibid.
C
More importantly, however, the Court’s “lowest common denominator” principle is inconsistent with the history of the canon of avoidance and is likely to have mischievous consequences. The modern canon of avoidance is a doctrine under which courts construe ambiguous statutes to avoid, constitutional doubts, but this doctrine has its origins in a very different form of the canon. Traditionally, the avoidance canon was not a doctrine under which courts read statutes to avoid mere constitutional doubts. Instead, it commanded courts, when faced with two plausible constructions of a statute — one constitutional and the other unconstitutional — to choose the constitutional reading.2 The traditional version of the canon thus requires courts to reach the issue whether the doubtful version of the statute is constitutional before adopting the construction that saves the statute from constitutional invalidity. A court faced with an ambiguous statute applies traditional avoidance by asking whether, given two plausible interpretations of that statute, one would be unconstitutional as applied to the plaintiff; and, if that interpretation is actually unconstitutional as applied to the plaintiff, the court picks the other (constitutional) read*396ing. The court does not inquire whether either of the interpretations would be unconstitutional if applied to third parties not before the court, unless the challenge is facial or otherwise implicates third-party rights.
This history suggests that the “lowest common denominator” principle is mistaken. Courts applying the modern version of the canon of avoidance should no more look to the rights of third parties than do courts using the traditional version. Under modern avoidance, in other words, an ambiguous statute should be read to avoid a constitutional doubt only if the statute is constitutionally doubtful as applied to the litigant before the court (again, unless the constitutional challenge involves third-party rights). Yet the Court’s lowest common denominator principle allows a limiting construction of an ambiguous statute prompted by constitutional doubts to infect other applications of the statute— even if the statute raises no constitutional doubt as applied to the specific litigant in a given case and even if the constitutionally unproblematic application of the statute to the litigant is severable from the constitutionally dubious applications. The lowest common denominator principle thus allows an end run around black-letter constitutional doctrine governing facial and as-applied constitutional challenges to statutes: A litigant ordinarily cannot attack statutes as constitutionally invalid based on constitutional doubts concerning other litigants or factual circumstances.
The Court misses the point by answering that the canon of constitutional avoidance “is not a method of adjudicating constitutional questions by other means,” and that the canon rests on a presumption that “Congress did not intend the alternative which raises serious constitutional doubts.” Ante, at 381. That is true, but in deciding whether a plausible interpretation “raises serious constitutional doubts,” a court must employ the usual rules of constitutional adju*397dication. See ante, at 380-381 (noting that whether an interpretation is constitutionally doubtful turns on whether it raises “a multitude of constitutional problems”); Zadvydas, 533 U. S., at 690-696 (extensively employing constitutional analysis). Those rules include doctrines governing third-party constitutional challenges and the like. Moreover, the reason that courts perform avoidance at all, in any form, is that we assume “Congress intends statutes to have effect to the full extent the Constitution allows.” United States v. Booker, ante, at 320 (Thomas, J., dissenting in part). Only my approach would extend § 1231(a)(6) to its full constitutional bound consistent with modern avoidance, by narrowing the statute on a case-by-case basis only if constitutional concerns are actually present. By contrast, under the majority’s lowest common denominator principle, a statute like § 1231(a)(6) must be narrowed once and for all based on constitutional concerns that may never materialize. In short, once narrowed in Zadvydas, § 1231(a)(6) now limits the Executive’s power to detain unadmitted aliens — even though indefinite detention of unadmitted aliens may be perfectly constitutional.
All of this shows why the sole support the majority offers for its lowest common denominator principle can be squared with my analysis. That support is a plurality opinion of this Court (reaffirmed by footnote dictum in Leocal v. Ashcroft, ante, at 11-12, n. 8), that stated that the rule of lenity applies to statutes so long as they have some criminal applications. Ante, at 380 (citing United States v. Thompson/Center Arms Co., 504 U. S. 505, 517 (1992)). To the extent that the rule of lenity is a constitutionally based clear statement rule, it is like vagueness doctrine, as its purpose is to ensure that those subjected to criminal prosecution have adequate notice of the conduct that the law prohibits. Cf., e. g., McBoyle v. United States, 283 U. S. 25, 27 (1931). Thompson/Center Arms is thus distinguishable, because our rules governing third-*398party challenges (rightly or wrongly) are more lenient in vagueness cases.3 Zadvydas, by contrast, was a straightforward as-applied constitutional challenge. It concerned a constitutional doubt that arose from § 1231(a)(6)’s application to the respondents there, not its hypothetical application to other aliens, as its careful distinction between admitted and inadmissible aliens shows. To the extent that the rule of lenity is a nonconstitutionally based presumption about the interpretation of criminal statutes, the Thompson!Center Arms interpretive principle is fundamentally different from the canon of constitutional avoidance, because the rule of lenity is wholly independent of the rules governing constitutional adjudication. Either way, this case does not support the majority’s restatement of modern avoidance principles.
The cases at bar illustrate well the exception to the normal operation of as-applied constitutional adjudication that the Court’s approach creates. Congress explicitly provided that unconstitutional applications of § 1281(a)(6) should be severed from constitutional applications.4 Congress has thus indicated that courts should examine whether § 1231(a)(6) raises a constitutional doubt application by application. After all, under the severability clause, if Zadvydas had held unconstitutional the indefinite detention of respondents Zadvydas and Ma, the constitutionality of the Secretary’s indefinite detention of Benitez and Martinez would remain an open question. Although Zadvydas did not formally hold § 1231(a)(6) to be unconstitutional as applied to the aliens before it, the same procedure should be followed when analyz*399ing whether § 1231(a)(6) raises a constitutional doubt.5 The Court today limits applications of § 1231(a)(6) that may well be constitutional solely on the basis of constitutional doubts as to other applications, and despite that the severability clause contemplates application-by-application examination of the statute’s constitutionality.
The Court misapprehends my interpretive approach. It suggests that I would “spare [us] the necessity of ever finding a statute unconstitutional as applied,” ante, at 384, and “would render every statute a chameleon, its meaning subject to change depending on the presence or absence of constitutional concerns in each individual case,” ante, at 382. My approach does none of this. I simply would read ambiguous statutes to avoid as-applied constitutional doubts only if those doubts are present in the ease before the Court. This leaves plenty of room for as-applied invalidation of statutes that are unambiguously unconstitutional. Nor would I permit a court to read every statute’s meaning to depend on constitutional concerns. That is permissible, in my view, only if the statute is ambiguous. Granted, I am thereby guilty of leaving courts free to interpret ambiguous statutes “as becoming inoperative when they ‘approach constitutional limits.’” Ante, at 384. That is hardly an absurd result— unless one considers the modern canon of constitutional *400avoidance itself to be absurd. Every application of that canon, by rejecting a plausible interpretation of a statute, reads the statute to be inoperative to the extent it raises a constitutional doubt or “limit.”
In truth, the Court’s aggressive application of modern constitutional avoidance doctrine poses the greater danger. A disturbing number of this Court’s cases have applied the canon of constitutional doubt to statutes that were on their face clear. See, e. g., INS v. St. Cyr, 533 U. S. 289, 327-336 (2001) (Scalia, J., dissenting); Public Citizen v. Department of Justice, 491 U. S. 440, 481-482 (1989) (Kennedy, J., concurring in judgment); Lowe v. SEC, 472 U. S. 181, 212-213 (1985) (White, J., concurring in result). This Court and others may now employ the “lowest common denominator” approach to limit the application of statutes wholesale by searching for hypothetical unconstitutional applications of them — or, worse yet, hypothetical constitutional doubts — despite the absence of any facial constitutional problem (at least, so long as those hypothetical doubts pose “a multitude of constitutional problems,” ante, at 380-381). This is so even if Congress has expressed its clear intent that unconstitutional applications should be severed from constitutional applications, regardless of whether the challenger has third-party standing to raise the constitutional issue, and without the need to engage in full-fledged constitutional analysis.
This danger is real. In St. Cyr, this Court held that the Immigration and Nationality Act (INA) did not divest district courts of jurisdiction under 28 U. S. C. § 2241 over ha-beas actions filed by criminal aliens to challenge removal orders, 533 U. S., at 314. The Court did so because it thought that otherwise the statute would preclude any avenue of judicial review of removal orders of criminal aliens, thus raising a serious Suspension Clause question. Id., at 305. This was a construction of (among other provisions) 8 U. S. C. §§ 1252(a)(1) and 1252(b)(9), and 28 U. S. C. § 2241, none of which distinguishes between criminal and noncriminal *401aliens. 533 U. S., at 308-314. The INA, however, clearly allows noncriminal aliens, unlike criminal aliens, a right to judicial review of removal decisions in the courts of appeals under the review provisions of § 1252(a)(1), and St. Cyr involved only criminal aliens. After St. Cyr, therefore, one would have thought that “noncriminal aliens seeking to challenge their removal orders . . . [would] still presumably be required to proceed directly to the court of appeals by way of petition for review, under the restrictive modified Hobbs Act review provisions set forth in § 1252(a)(1),” rather than sue directly under the habeas statute. Id., at 335 (Scalia, J., joined by Rehnquist, C. J., and O’Connor and Thomas, JJ., dissenting). Yet lower courts, relying on a version of the Court’s “lowest common denominator” principle, have held just the opposite: They have entertained noncriminal aliens’ habeas actions challenging removal orders. Chmakov v. Blackman, 266 F. 3d 210, 214-215 (CA3 2001); see also Riley v. INS, 310 F. 3d 1253, 1256 (CA10 2002); Liu v. INS, 293 F. 3d 36, 38-41 (CA2 2002). The logic in allowing noncriminal aliens, who have a right to judicial review of removal decisions, to take advantage of constitutional doubt that arises from precluding any avenue of judicial review for criminal aliens, see St. Cyr, supra, at 305, escapes me.
II
The Court is also mistaken in affording Zadvydas stare decisis effect. Zadvydas was wrong in both its statutory and its constitutional analysis for the reasons expressed well by the dissents in that case. See 533 U. S., at 705-718 (opinion of Kennedy, J.); id., at 702-705 (opinion of Scalia, J.). I continue to adhere to those views and will not repeat the analysis of my colleagues. I write only to explain why I do not consider Zadvydas to bind us.
Zadvydas cast itself as a statutory case, but that fact should not prevent us from overruling it. It is true that we give stronger stare decisis effect to our holdings in statutory. *402cases than in constitutional cases. See, e. g., Hilton v. South Carolina Public Railways Comm’n, 502 U. S. 197, 205 (1991). This rule, however, is not absolute, and we should not hesitate to allow our precedent to yield to the true meaning of an Act of Congress when our statutory precedent is “unworkable” or “badly reasoned.” Holder v. Hall, 512 U. S. 874, 936 (1994) (Thomas, J., concurring in judgment) (quoting Payne v. Tennessee, 501 U. S. 808, 827 (1991); internal quotation marks omitted). “[W]e have never applied stare decisis mechanically to prohibit overruling our earlier decisions determining the meaning of statutes.” Monell v. New York City Dept. of Social Servs., 436 U. S. 658, 695 (1978). The mere fact that Congress can overturn our cases by statute is no excuse for failing to overrule a statutory precedent of ours that is clearly wrong, for the realities of the legislative process often preclude readopting the original meaning of a statute that we have upset.
Zadvydas’ reading of § 1231(a)(6) is untenable. Section 1231(a)(6) provides that aliens whom the Secretary of Homeland Security has ordered removed “may be detained beyond the removal period.” There is no qualification to this authorization, and no reference to a “reasonable time” limitation. Just as we exhaust the aid of the “traditional tools of statutory construction,” Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 843, n. 9 (1984), before deferring to an agency’s interpretation of a statute, so too should we exhaust those tools before deciding that a statute is ambiguous and that an alternative plausible construction of the statute should be adopted.
Application of those traditional tools begins and ends with the text of § 1231(a)(6). Zadvydas’ observation that “if Congress had meant to authorize long-term detention of unre-movable aliens, it certainly could have spoken in clearer terms,” 533 U. S., at 697, proves nothing. Congress could have spoken more clearly in any statutory case in which the statute does not mention the particular factual scenario be*403fore the Court. Congress provided for a “reasonable time” limit to detentions pending removal in other portions of § 1231. Id., at 708 (Kennedy, J., dissenting). Its failure to do the same in § 1231(a)(6) confirms what is unmistakable from its terms: that there is no time limit on the Secretary’s power to detain aliens. There is no textually evident alternative construction that would avoid the constitutional doubts identified by the majority.
Even apart from the Court’s incredible reading of § 1231(a)(6), the normal reason for affording our statutory holdings strong stare decisis effect — that Congress is free to overrule them if it disagrees — does not apply to Zadvydas. Zadvydas is a statutory case in name only. Although the Zadvydas majority purported to find indefinite detention only constitutionally doubtful, its lengthy analysis strongly signaled to Congress that indefinite detention of admitted aliens would be unconstitutional. Indeed, far from avoiding that constitutional question in Zadvydas, the Court took it head on, giving it extended treatment. Id., at 690-697; but see ante, at 381 (noting the “fundamenta[l]” tenet that “[t]he canon [of constitutional avoidance] is not a method of adjudicating constitutional questions by other means”). Zadvydas makes clear that the Court thought indefinite detention to be more than constitutionally suspect, and there is evidence that some Members of Congress understood as much.6 This is why the Court’s assurance that if “the security of our bor*404ders will be compromised if [the United States] must release into the country inadmissible aliens who cannot be removed[,] Congress can attend to it,” ante, at 386, rings hollow. Short of constitutional amendment, it is only within the power of this Court to correct Zadvydas’ error.
The Court points to 8 U. S. C. § 1226a(a)(6) (2000 ed., Supp. II), a statute that Congress passed shortly after Zadvydas, as evidence that Congress can correct Zadvydas’ mistake. Ante, at 386, n. 8. This statute only confirms my concern that Zadvydas is legislatively uncorrectable. Section 1226a(a)(6) authorizes detention for a period of six months beyond the removal period of aliens who present a national security threat, but only to the extent that those aliens’ removal is not reasonably foreseeable. Ante, at 386, n. 8. Yet Zadvydas conceded that indefinite detention might not violate due process in “certain special and narrow nonpunitive circumstances . . . where a special justification, such as harm-threatening mental illness, outweighs the individual’s constitutionally protected interest in avoiding physical restraint.” 533 U. S., at 690 (internal quotation marks omitted). Moreover, Zadvydas set a 6-month presumptive outer limit on the detention power. Id., at 701. Congress crafted § 1226a(a)(6) to operate within the boundaries Zadvydas set. This provision says nothing about whether Congress may authorize detention of aliens for greater lengths of time or for reasons the Court found constitutionally problematic in Zadvydas.
* * *
For the foregoing reasons, I would affirm the judgment of the Eleventh Circuit and reverse the judgment of the Ninth Circuit. I therefore respectfully dissent.
Section 1367(d) provides that “[t]he period of limitations for any claim asserted under [§ 1367(a)], and for any other claim in the same action that is- voluntarily dismissed at the same time as or after the dismissal of the claim under [§ 1367(a)], shall be tolled while the claim is pending and for a period of 30 days after it is dismissed unless State law provides for a longer tolling period.”
See Rust v. Sullivan, 500 U. S. 173, 190-191 (1991) (distinguishing the classic and modern versions of the canon and citing cases); Hooper v. California, 155 U. S. 648, 657 (1895) (“The elementary rule is that every reasonable construction must be resorted to, in order to save a statute from unconstitutionality”); Mossman v. Higginson, 4 Dall. 12, 14 (1800) (reasoning that the statute under review “can, and must receive a construction, consistent with the constitution”); Ex parte Randolph, 20 F. Cas. 242, 254 (No. 11,558) (CC Va. 1833) (Marshall, J.); Vermeule, Saving Constructions, 85 Geo. L. J. 1945, 1949 (1997); H. Black, Handbook on the Construction and Interpretation of the Laws 113-114 (2d ed. 1911). The modern version seems to have originated in United States ex rel. Attorney General v. Delaware & Hudson Co., 213 U. S. 366, 408 (1909).
See, e. g., Chicago v. Morales, 527 U. S. 41, 55, and n. 22 (1999) (plurality opinion); Kolender v. Lawson, 461 U. S. 352, 358-359, n. 8 (1983); Papachristou v. Jacksonville, 405 U. S. 156 (1972).
“If any provision of this division... or the application of such provision to any person or circumstances is held to be unconstitutional, the remainder of this division and the application of the provisions of this division to any person or circumstance shall not be affected thereby.” Note following 8 U. S. C. § 1101, p. 840 (separability).
Crowell v. Benson, 285 U. S. 22 (1932), bolsters my approach. Employing the canon of avoidance, the Court construed a statute in that case to allow judicial review of jurisdictional facts but not legislative facts. It did so even though the terms of the statute itself did not distinguish between the two sorts of facts. . Id., at 62-63. The presence of a severability provision in the statute gave “assurance that there [was] no violation of the purpose of the Congress in sustaining the determinations of fact of the deputy commissioner where he acts within his authority in passing upon compensation claims while denying finality to his conclusions as to the jurisdictional facts upon which the valid application of the statute depends.” Ibid. So too here, the presence of a severability provision should reassure the Court that applying Zadvydas’ limiting construction of § 1231(a)(6) to some aliens and not others is consistent with the statute.
See H. R. Conf. Rep. No. 108-10, p. 600 (2003) (“A recent Supreme Court decision held that criminal aliens cannot be detained indefinitely,” no doubt referring to Zadvydas); H. R. Rep. No. 108-724, pt. 5, p. 191 (2004) (“The danger posed by the requirement that these aliens be allowed to remain in the U. S. was increased exponentially by the 2001 Supreme Court decision of Zadvydas v. Davis, in which the Court made clear that it would strike down as unconstitutional the indefinite detention by [the Secretary] of aliens with removal orders whose countries will not take them back, except in the most narrow of circumstances” (footnote omitted)); 147 Cong. Rec. 20729 (2001) (“Indefinite detention of aliens is permitted only in extraordinary circumstances,” citing Zadvydas).