Justice Scalia
delivered the opinion of the Court.
The question presented in this case is whether § 27A(b) of the Securities Exchange Act of 1934, to the extent that it requires federal courts to reopen final judgments in private civil actions under § 10(b) of the Act, contravenes the Constitution’s separation of powers or the Due Process Clause of the Fifth Amendment.
I
In 1987, petitioners brought a civil action against respondents in the United States District Court for the Eastern District of Kentucky. The complaint alleged that in 1983 and 1984 respondents had committed fraud and deceit in the sale of stock in violation of § 10(b) of the Securities Exchange Act of 1934 and Rule 10b-5 of the Securities and Exchange Commission. The case was mired in pretrial proceedings in the District Court until June 20, 1991, when we decided Lampf, Pleva, Lipkind, Prupis & Petigrow v. Gilbertson, 501 U. S. 350. Lampf held that “[litigation instituted pursuant to § 10(b) and Rule 10b-5 . . . must be commenced within one year after the discovery of the facts constituting the violation and within three years after such violation.” Id., at [214]*214364. We applied that holding to the plaintiff-respondents in Lampf itself, found their suit untimely, and reinstated a summary judgment previously entered in favor of the defendant-petitioners. Ibid. On the same day we decided James B. Beam Distilling Co. v. Georgia, 501 U. S. 529 (1991), in which a majority of the Court held, albeit in different opinions, that a new rule of federal law that is applied to the parties in the case announcing the rule must be applied as well to all cases pending on direct review. See Harper v. Virginia Dept. of Taxation, 509 U. S. 86, 92 (1993). The joint effect of Lampf and Beam was to mandate application of the l-year/3-year limitations period to petitioners’ suit. The District Court, finding that petitioners’ claims were untimely under the Lampf rule, dismissed their action with prejudice on August 13, 1991. Petitioners filed no appeal; the judgment accordingly became final 30 days later. See 28 U. S. C. § 2107(a) (1988 ed., Supp. V); Griffith v. Kentucky, 479 U. S. 314, 321, n. 6 (1987).
On December 19, 1991, the President signed the Federal Deposit Insurance Corporation Improvement Act of 1991, 105 Stat. 2236. Section 476 of the Act — a section that had nothing to do with FDIC improvements — became §27A of the Securities Exchange Act of 1934, and was later codified as 15 U. S. C. § 78aa-1 (1988 ed., Supp. V). It provides:
“(a) Effect on pending causes of action
“The limitation period for any private civil action implied under section 78j(b) of this title [§ 10(b) of the Securities Exchange Act of 1934] that was commenced on or before June 19, 1991, shall be the limitation period provided by the laws applicable in the jurisdiction, including principles of retroactivity, as such laws existed on June 19,1991.
“(b) Effect on dismissed causes of action
“Any private civil action implied under section 78j(b) of this title that was commenced on or before June 19, 1991—
[215]*215“(1) which was dismissed as time barred subsequent to June 19,1991, and
“(2) which would have been timely filed under the limitation period provided by the laws applicable in the jurisdiction, including principles of retroactivity, as such laws existed on June 19,1991,
“shall be reinstated on motion by the plaintiff not later than 60 days after December 19, 1991.”
On February 11, 1992, petitioners returned to the District Court and filed a motion to reinstate the action previously dismissed with prejudice. The District Court found that the conditions set out in §§27A(b)(l) and (2) were met, so that petitioners’ motion was required to be granted by the terms of the statute. It nonetheless denied the motion, agreeing with respondents that §27A(b) is unconstitutional. Memorandum Opinion and Order, Civ. Action No. 87-438 (ED Ky., Apr. 13, 1992). The United States Court of Appeals for the Sixth Circuit affirmed. 1 F. 3d 1487 (1993). We granted certiorari. 511 U. S. 1141 (1994).1
II
Respondents bravely contend that §27A(b) does not require federal courts to reopen final judgments, arguing first that the reference to “the laws applicable in the jurisdiction . . . as such laws existed on June 19, 1991” (the day before Lamp† was decided) may reasonably be construed to refer precisely to the limitations period provided in Lampf itself, in which case petitioners’ action was time barred even under [216]*216§ 27A.2 It is true that “[a] judicial construction of a statute is an authoritative statement of what the statute meant before as well as after the decision of the case giving rise to that construction.” Rivers v. Roadway Express, Inc., 511 U. S. 298, 312-313 (1994); see also id., at 313, n. 12. But respondents’ argument confuses the question of what the law in fact was on June 19, 1991, with the distinct question of what § 27A means by its reference to what the law was. We think it entirely clear that it does not mean the law enunciated in Lampf, for two independent reasons. First, Lampf provides a uniform, national statute of limitations (instead of using the applicable state limitations period, as lower federal courts had previously done. See Lampf, 501 U. S., at 354, and n. 1). If the statute referred to that law, its reference to the “laws applicable in the jurisdiction” (emphasis added) would be quite inexplicable. Second, if the statute refers to the law enunciated in Lampf, it is utterly without effect, a result to be , avoided if possible. American Nat. Red Cross v. S. G., 505 U. S. 247, 263-264 (1992); see 2A N. Singer, Sutherland on Statutory Construction § 46.06 (Sands rev. 4th ed. 1984). It would say, in subsection (a), that the limitations period is what the Supreme Court has held to be the limitations period; and in subsection (b), that suits dismissed as untimely under Lampf which were timely under Lampf (a null set) shall be reinstated. To avoid a constitutional question by holding that Congress enacted, and the President approved, a blank sheet of paper would indeed constitute “disingenuous evasion.” George Moore Ice Cream Co. v. Rose, 289 U. S. 373, 379 (1933).
[217]*217As an alternative reason why § 27A(b) does not require the reopening of final judgments, respondents suggest that the subsection applies only to cases still pending in the federal courts when §27A was enacted. This has only half the defect of the first argument, for it makes only half of §27A purposeless — §27A(b). There is no need to “reinstate” actions that are still pending; § 27A(a) (the new statute of limitations) could and would be applied by the courts of appeals. On respondents’ reading, the only consequence of §27A(b) would be the negligible one of permitting the plaintiff in the pending appeal from a statute-of-limitations dismissal to return immediately to the district court, instead of waiting for the court of appeals’ reversal. To enable § 27A(b) to achieve such an insignificant consequence, one must disregard the language of the provision, which refers generally to suits “dismissed as time barred.” It is perhaps arguable that this does not include suits that are not yet finally dismissed, i. e., suits still pending on appeal; but there is no basis for the contention that it includes only those. In short, there is no reasonable construction on which §27A(b) does not require federal courts to reopen final judgments in suits dismissed with prejudice by virtue of Lampf.
Ill
Respondents submit that § 27A(b) violates both the separation of powers and the Due Process Clause of the Fifth Amendment.3 Because the latter submission, if correct, might dictate a similar result in a challenge to state legislation under the Fourteenth Amendment, the former is the narrower ground for adjudication of the constitutional questions in the case, and we therefore consider it first. Ashwander v. TVA, 297 U. S. 288, 347 (1936) (Brandéis, J., concurring). We conclude that in § 27A(b) Congress has exceeded its authority by requiring the federal courts to exercise. [218]*218“[t]he judicial Power of the United States,” U. S. Const., Art. III, §1, in a manner repugnant to the text, structure, and traditions of Article III.
Our decisions to date have identified two types of legislation that require federal courts to exercise the judicial power in a manner that Article III forbids. The first appears in United States v. Klein, 13 Wall. 128 (1872), where we refused to give effect to a statute that was said “[to] prescribe rules of decision to the Judicial Department of the government in cases pending before it.” Id., at 146. Whatever the precise scope of Klein, however, later decisions have made clear that its prohibition does not take hold when Congress “amend[s] applicable law.” Robertson v. Seattle Audubon Soc., 503 U. S. 429, 441 (1992). Section 27A(b) indisputably does set out substantive legal standards for the Judiciary to apply, and in that sense changes the law (even if solely retroactively). The second type of unconstitutional restriction upon the exercise of judicial power identified by past cases is exemplified by Hayburn’s Case, 2 Dall. 409 (1792), which stands for the principle that Congress cannot vest review of the decisions of Article III courts in officials of the Executive Branch. See, e. g., Chicago & Southern Air Lines, Inc. v. Waterman S. S. Corp., 333 U. S. 103 (1948). Yet under any application of §27A(b) only courts are involved; no officials of other departments sit in direct review of their decisions. Section 27A(b) therefore offends neither of these previously established prohibitions.
We think, however, that § 27A(b) offends a postulate of Article III just as deeply rooted in our law as those we have mentioned. Article III establishes a “judicial department” with the “province and duty ... to say what the law is” in particular cases and controversies. Marbury v. Madison, 1 Cranch 137, 177 (1803). The record of history shows that the Framers crafted this charter of the judicial department with an expressed understanding that it gives the Federal Judiciary the power, not merely to rule on cases, but to de[219]*219tide them, subject to review only by superior courts in the Article III hierarchy — with an understanding, in short, that “a judgment conclusively resolves the case” because “a ‘judicial Power’ is one to render dispositive judgments.” Easter-brook, Presidential Review, 40 Case W. Res. L. Rev. 905, 926 (1990). By retroactively commanding the federal courts to reopen final judgments, Congress has violated this fundamental principle.
A
The Framers of our Constitution lived among the ruins of a system of intermingled legislative and judicial powers, which had been prevalent in the colonies long before the Revolution, and which after the Revolution had produced factional strife and partisan oppression. In the 17th and 18th centuries colonial assemblies and legislatures functioned as courts of equity of last resort, hearing original actions or providing appellate review of judicial judgments. G. Wood, The Creation of the American Republic 1776-1787, pp. 154-155 (1969). Often, however, they chose to correct the judicial process through special bills or other enacted legislation. It was common for such legislation not to prescribe a resolution of the dispute, but rather simply to set aside the judgment and order a new trial or appeal. M. Clarke, Parliamentary Privilege in the American Colonies 49-51 (1943). See, e.g., Judicial Action by the Provincial Legislature of Massachusetts, 15 Harv. L. Rev. 208 (1902) (collecting documents from 1708-1709); 5 Laws of New Hampshire, Including Public and Private Acts, Resolves, Votes, Etc., 1784-1792 (Metcalf ed. 1916). Thus, as described in our discussion of Hayburn’s Case, supra, at 218, such legislation bears not on the problem of interbranch review but on the problem of finality of judicial judgments.
The vigorous, indeed often radical, populism of the revolutionary legislatures and assemblies increased the frequency of legislative correction of judgments. Wood, supra, at 155-156, 407-408. See also INS v. Chadha, 462 U. S. 919, 961 [220]*220(1983) (Powell, J., concurring). “The period 1780-1787 . . . was a period of ‘constitutional reaction’ ” to these developments, “which ... leaped suddenly to its climax in the Philadelphia Convention.” E. Corwin, The Doctrine of Judicial Review 37 (1914). Voices from many quarters, official as well as private, decried the increasing legislative interference with the private-law judgments of the courts. In 1786, the Vermont Council of Censors issued an “Address of the Council of Censors to the Freemen of the State of Vermont” to fulfill the council’s duty, under the State Constitution of 1784, to report to the people “ ‘whether the legislative and executive branches of government have assumed to themselves, or exercised, other or greater powers than they are entitled to by the Constitution.’” Vermont State Papers 1779-1786, pp. 531, 533 (Slade ed. 1823). A principal method of usurpation identified by the censors was “[t]he instances ... of judgments being vacated by legislative acts.” Id., at 540. The council delivered an opinion
“that the General Assembly, in all the instances where they have vacated judgments, recovered in due course of law, (except where the particular circumstances of the case evidently made it necessary to grant a new trial) have exercised a power not delegated, or intended to be delegated, to them, by the Constitution. ... It super-cedes the necessity of any other law than the pleasure of the Assembly, and of any other court than themselves: for it is an imposition on the suitor, to give him the trouble of obtaining, after several expensive trials, a final judgment agreeably to the known established laws of the land; if the Legislature, by a sovereign act, can interfere, reverse the judgment, and decree in such manner, as they, unfettered by rules, shall think proper.” Ibid.
So too, the famous report of the Pennsylvania Council of Censors in 1784 detailed the abuses of legislative interference with the courts at the behest of private interests and [221]*221factions. As the General Assembly had (they wrote) made a custom of “extending their deliberations to the cases of individuals,” the people had “been taught to consider an application to the legislature, as a shorter and more certain mode of obtaining relief from hardships and losses, than the usual process of law.” The censors noted that because “favour and partiality have, from the nature of public bodies of men, predominated in the distribution of this relief... [t]hese dangerous procedures have been too often recurred to, since the revolution.” Report of the Committee of the Council of Censors 6 (Bailey ed. 1784).
This sense of a sharp necessity to separate the legislative from the judicial power, prompted by the crescendo of legislative interference with private judgments of the courts, triumphed among the Framers of the new Federal Constitution. See Corwin, The Progress of Constitutional Theory Between the Declaration of Independence and the Meeting of the Philadelphia Convention, 30 Am. Hist. Rev. 511, 514-517 (1925). The Convention made the critical decision to establish a judicial department independent of the Legislative Branch by providing that “the judicial Power of the United States shall be vested in one supreme Court, and in such inferior Courts as the Congress may from time to time ordain and establish.” Before and during the debates on ratification, Madison, Jefferson, and Hamilton each wrote of the factional disorders and disarray that the system of legislative equity had produced in the years before the framing; and each thought that the separation of the legislative from the judicial power in the new Constitution would cure them. Madison’s Federalist No. 48, the famous description of the process by which “[tjhe legislative department is every where extending the sphere of its activity, and drawing all power into its impetuous vortex,” referred to the report of the Pennsylvania Council of Censors to show that in that State “cases belonging to the judiciary department [had been] frequently drawn within legislative cognizance and determination.” The Fed[222]*222eralist No. 48, pp. 333, 337 (J. Cooke ed. 1961). Madison relied as well on Jefferson’s Notes on the State of Virginia, which mentioned, as one example of the dangerous concentration of governmental powers into the hands of the legislature, that “the Legislature ... in many instances decided rights which should have been left to judiciary controversy.” Id., at 336 (emphasis deleted).4
If the need for separation of legislative from judicial power was plain, the principal effect to be accomplished by that separation was even plainer. As Hamilton wrote in his exegesis of Article III, § 1, in The Federalist No. 81:
“It is not true . .. that the parliament of Great Britain, or the legislatures of the particular states, can rectify the exceptionable decisions of their respective courts, in any other sense than might be done by a future legislature of the United States. The theory neither of the British, nor the state constitutions, authorises the revisal of a judicial sentence, by a legislative act. ... A legislature without exceeding its province cannot reverse a determination once made, in a particular case; though it may prescribe a new rule for future cases.” The Federalist No. 81, p. 545 (J. Cooke ed. 1961).
The essential balance created by this allocation of authority was a simple one. The Legislature would be possessed of power to “prescribe] the rules by which the duties and rights of every citizen are to be regulated,” but the power of “[t]he interpretation of the laws” would be “the proper and peculiar province of the courts.” Id., No. 78, at 523, 525. [223]*223See also Corwin, The Doctrine of Judicial Review, at 42. The Judiciary would be, “from the nature of its functions,... the [department] least dangerous to the political rights of the constitution,” not because its acts were subject to legislative correction, but because the binding effect of its acts was limited to particular cases and controversies. Thus, “though individual oppression may now and then proceed from the courts of justice, the general liberty of the people can never be endangered from that quarter:... so long as the judiciary remains truly distinct from both the legislative and executive.” The Federalist No. 78, at 522, 523.
Judicial decisions in the period immediately after ratification of the Constitution confirm the understanding that it forbade interference with the final judgments of courts. In Calder v. Bull, 3 Dall. 386 (1798), the Legislature of Connecticut had enacted a statute that set aside the final judgment of a state court in a civil case. Although the issue before this Court was the construction of the Ex Post Facto Clause, Art. I, §10, Justice Iredell (a leading Federalist who had guided the Constitution to ratification in North Carolina) noted that
“the Legislature of [Connecticut] has been in the uniform, uninterrupted, habit of exercising a general superintending power over its courts of law, by granting new trials. It may, indeed, appear strange to some of us, that in any form, there should exist a power to grant, with respect to suits depending or adjudged, new rights of trial, new privileges of proceeding, not previously recognized and regulated by positive institutions____ The power ... is judicial in its nature; and whenever it is exercised, as in the present instance, it is an exercise of judicial, not of legislative, authority.” Id., at 398.
The state courts of the era showed a similar understanding of the separation of powers, in decisions that drew little distinction between the federal and state constitutions. To [224]*224choose one representative example from a multitude: In Bates v. Kimball, 2 Chipman 77 (Vt. 1824), a special Act of the Vermont Legislature authorized a party to appeal from the judgment of a court even though, under the general law, the time for appeal had expired. The court, noting that the unappealed judgment had become final, set itself the question “Have the Legislature power to vacate or annul an existing judgment between party and party?” Id., at 83. The answer was emphatic: “The necessity of a distinct and separate existence of the three great departments of government . . . had been proclaimed and enforced by . . . Blackstone, Jefferson and Madison,” and had been “sanctioned by the people of the United States, by being adopted in terms more or less explicit, into all their written constitutions.” Id., at 84. The power to annul a final judgment, the court held (citing Hayburn’s Case, 2 Dall., at 410), was “an assumption of Judicial power,” and therefore forbidden. Bates v. Kimball, supra, at 90. For other examples, see Merrill v. Sherburne, 1 N. H. 199 (1818) (legislature may not vacate a final judgment and grant a new trial); Lewis v. Webb, 3 Greenleaf 299 (Me. 1825) (same); T. Cooley, Constitutional Limitations 95-96 (1868) (collecting cases); J. Sutherland, Statutory Construction 18-19 (J. Lewis ed. 1904) (same).
By the middle of the 19th century, the constitutional equilibrium created by the separation of the legislative power to make general law from the judicial power to apply that law in particular cases was so well understood and accepted that it could survive even Dred Scott v. Sandford, 19 How. 393 (1857). In his First Inaugural Address, President Lincoln explained why the political branches could not, and need not, interfere with even that infamous judgment:
“I do not forget the position assumed by some, that constitutional questions are to be decided by the Supreme Court; nor do I deny that such decisions must be binding in any case, upon the parties to a suit, as to the object of that suit.... And while it is obviously possible that [225]*225such decision may be erroneous in any given case, still the evil effect following it, being limited to that particular case, with the chance that it may be over-ruled, and never become a precedent for other cases, can better be borne than could the evils of a different practice.” 4 R. Basler, The Collected Works of Abraham Lincoln 268 (1953) (First Inaugural Address 1861).
And the great constitutional scholar Thomas Cooley addressed precisely the question before us in his 1868 treatise:
“If the legislature cannot thus indirectly control the action of the courts, by requiring of them a construction of the law according to its own views, it is very plain it cannot do so directly, by setting aside their judgments, compelling them to grant new trials, ordering the discharge of offenders, or directing what particular steps shall be taken in the progress of a judicial inquiry.” Cooley, supra, at 94-95.
B
Section 27A(b) effects a clear violation of the separation-of-powers principle we have just discussed. It is, of course, retroactive legislation, that is, legislation that prescribes what the law was at an earlier time, when the act whose effect is controlled by the legislation occurred — in this case, the filing of the initial Rule 10b-5 action in the District Court. When retroactive legislation requires its own application in a case already finally adjudicated, it does no more and no less than “reverse a determination once made, in a particular case.” The Federalist No. 81, at 545. Our decisions stemming from Hayburn’s Case — although their precise holdings are not strictly applicable here, see supra, at 218 — have uniformly provided fair warning that such an act exceeds the powers of Congress. See, e.g., Chicago & Southern Air Lines, Inc., 333 U. S., at 113 (“Judgments within the powers vested in courts by the Judiciary Article [226]*226of the Constitution may not lawfully be revised, overturned or refused faith and credit by another Department of Government”); United States v. O’Grady, 22 Wall. 641, 647-648 (1875) (“Judicial jurisdiction implies the power to hear and determine a cause, and ... Congress cannot subject the judgments of the Supreme Court to the re-examination and revision of any other tribunal”); Gordon v. United States, 117 U. S. Appx. 697, 700-704 (1864) (opinion of Taney, C. J.), (judgments of Article III courts are “final and conclusive1 upon the rights of the parties”); Hayburn’s Case, 2 Dall., at 411 (opinion of Wilson and Blair, JJ., and Peters, D. J.) (“[RJevision and control” of Article III judgments is “radically inconsistent with the independence of that judicial power which is vested in the courts”); id., at 413 (opinion of Iredell, J., and Sitgreaves, D. J.) (“[N]o decision of any court of the United States can, under any circumstances, ... be liable to a revision, or even suspension, by the [legislature itself, in whom no judicial power of any kind appears to be vested”). See also Pennsylvania v. Wheeling & Belmont Bridge Co., 18 How. 421, 431 (1856) (“[I]t is urged, that the act of congress cannot have the effect and operation to annul the judgment of the court already rendered, or the rights determined thereby .... This, as a general proposition, is certainly not to be denied, especially as it respects adjudication upon the private rights of parties. When they have passed into judgment the right becomes absolute, and it is the duty of the court to enforce it”). Today those clear statements must either be honored, or else proved false.
It is true, as petitioners contend, that Congress can always revise the judgments of Article III courts in one sense: When a new law makes clear that it is retroactive, an appellate court must apply that law in reviewing judgments still on appeal that were rendered before the law was enacted, and must alter the outcome accordingly. See United States v. Schooner Peggy, 1 Cranch 103 (1801); Landgraf v. US I Film Products, 511 U. S. 244, 273-280 (1994). Since that is [227]*227so, petitioners argue, federal courts must apply the “new” law created by §27A(b) in finally adjudicated cases as well; for the line that separates lower court judgments that are pending on appeal (or may still be appealed), from lower court judgments that are final, is determined by statute, see, e. g., 28 U. S. C. § 2107(a) (30-day time limit for appeal to federal court of appeals), and so cannot possibly be a constitutional line. But a distinction between judgments from which all appeals have been forgone or completed, and judgments that remain on appeal (or subject to being appealed), is implicit in what Article III creates: not a batch of unconnected courts, but a judicial department composed of “inferior Courts” and “one supreme Court.” Within that hierarchy, the decision of an inferior court is not (unless the time for appeal has expired) the final word of the department as a whole. It is the obligation of the last court in the hierarchy that rules on the case to give effect to Congress’s latest enactment, even when that has the effect of overturning the judgment of an inferior court, since each court, at every level, must “decide according to existing laws.” Schooner Peggy, supra, at 109. Having achieved finality, however, a judicial decision becomes the last word of the judicial department with regard to a particular case or controversy, and Congress may not declare by retroactive legislation that the law applicable to that very case was something other than what the courts said it was. Finality of a legal judgment is determined by statute, just as entitlement to a government benefit is a statutory creation; but that no more deprives the former of its constitutional significance for separation-of-powers analysis than it deprives the latter of its significance for due process purposes. See, e. g., Cleveland Bd. of Ed. v. Loudermill, 470 U. S. 532 (1985); Meachum v. Fano, 427 U. S. 215 (1976).
To be sure, § 27A(b) reopens (or directs the reopening of) final judgments in a whole class of cases rather than in a particular suit. We do not see how that makes any differ[228]*228ence. The separation-of-powers violation here, if there is any, consists of depriving judicial judgments of the conclusive effect that they had when they were announced, not of acting in a manner — viz., with particular rather than general effect — that is unusual (though, we must note, not impossible) for a legislature. To be sure, a general statute such as this one may reduce the perception that legislative interference with judicial judgments was prompted by individual favoritism; but it is legislative interference with judicial judgments nonetheless. Not favoritism, nor even corruption, but power is the object of the separation-of-powers prohibition. The prohibition is violated when an individual final judgment is legislatively rescinded for even the very best of reasons, such as the legislature’s genuine conviction (supported by all the law professors in the land) that the judgment was wrong; and it is violated 40 times over when 40 final judgments are legislatively dissolved.
It is irrelevant as well that the final judgments reopened by § 27A(b) rested on the bar of a statute of limitations. The rules of finality, both statutory and judge made, treat a dismissal on statute-of-limitations grounds the same way they treat a dismissal for failure to state a claim, for failure to prove substantive liability, or for failure to prosecute: as a judgment on the merits. See, e.g., Fed. Rule Civ. Proc. 41(b); United States v. Oppenheimer, 242 U. S. 85, 87-88 (1916). Petitioners suggest, directly or by implication, two reasons why a merits judgment based on this particular ground may be uniquely subject to congressional nullification. First, there is the fact that the length and indeed even the very existence of a statute of limitations upon a federal cause of action is entirely subject to congressional control. But virtually all of the reasons why a final judgment on the merits is rendered on a federal claim are subject to congressional control. Congress can eliminate, for example, a particular element of a cause of action that plaintiffs have found it difficult to establish; or an evidentiary rule that has often [229]*229excluded essential testimony; or a rule of offsetting wrong (such as contributory negligence) that has often prevented recovery. To distinguish statutes of limitations on the ground that they are mere creatures of Congress is to distinguish them not at all. The second supposedly distinguishing characteristic of a statute of limitations is that it can be extended, without violating the Due Process Clause, after the cause of the action arose and even after the statute itself has expired. See, e. g., Chase Securities Corp. v. Donaldson, 325 U. S. 304 (1945). But that also does not set statutes of limitations apart. To mention only one other broad category of judgment-producing legal rule: Rules of pleading and proof can similarly be altered after the cause of action arises, Landgraf v. USI Film Products, supra, at 275, and n. 29, and even, if the statute clearly so requires, after they have been applied in a case but before final judgment has been entered. Petitioners’ principle would therefore lead to the conclusion that final judgments rendered on the basis of a stringent (or, alternatively, liberal) rule of pleading or proof may be set aside for retrial under a new liberal (or, alternatively, stringent) rule of pleading or proof. This alone provides massive scope for undoing final judgments and would substantially subvert the doctrine of separation of powers.
The central theme of the dissent is a variant on these arguments. The dissent maintains that Lampf “announced” a new statute of limitations, post, at 246, in an act of “judicial ... lawmaking,” post, at 247, that “changed the law,” post, at 250. That statement, even if relevant, would be wrong. The point decided in Lampf had never before been addressed by this Court, and was therefore an open question, no matter what the lower courts had held at the time. But the more important point is that Lampf as such is irrelevant to this case. The dissent itself perceives that “[w]e would have the same issue to decide had Congress enacted the Lampf rule,” and that the Lampf rule’s genesis. in judicial lawmaking rather than, shall we say, legislative lawmaking, “should not [230]*230affect the separation-of-powers analysis.” Post, at 247. Just so. The issue here is not the validity or even the source of the legal rule that produced the Article III judgments, but rather the immunity from legislative abrogation of those judgments themselves. The separation-of-powers question before us has nothing to do with Lanvpf, and the dissent’s attack on Lampf has nothing to do with the question before us.
C
Apart from the statute we review today, we know of no instance in which Congress has attempted to set aside the final judgment of an Article III court by retroactive legislation. That prolonged reticence would be amazing if such interference were not understood to be constitutionally proscribed. The closest analogue that the Government has been able to put forward is the statute at issue in United States v. Sioux Nation, 448 U. S. 371 (1980). That law required the Court of Claims, “ ‘[notwithstanding any other provision of law . . . [to] review on the merits, without regard to the defense of res judicata or collateral estoppel,’ ” a Sioux claim for just compensation from the United States— even though the Court of Claims had previously heard and rejected that very claim. We considered and rejected separation-of-powers objections to the statute based upon Hayburn’s Case and United States v. Klein. See 448 U. S., at 391-392. The basis for our rejection was a line of precedent (starting with Cherokee Nation v. United States, 270 U. S. 476 (1926)) that stood, we said, for the proposition that “Congress has the power to waive the res judicata effect of a prior judgment entered in the Government’s favor on a claim against the United States.” Sioux Nation, 448 U. S., at 397. And our holding was as narrow as the precedent on which we had relied: “In sum,. . . Congress’ mere waiver of the res judicata effect of a prior judicial decision rejecting the validity of a legal claim against the United States does [231]*231not violate the doctrine of separation of powers.” Id., at 407.5
The Solicitor General suggests that even if Sioux Nation is read in accord with its holding, it nonetheless establishes that Congress may require Article III courts to reopen their final judgments, since “if res judicata were compelled by Article III to safeguard the structural independence of the courts, the doctrine would not be subject to waiver by any party litigant.” Brief for United States 27 (citing Commodity Futures Trading Comm’n v. Schor, 478 U. S. 833,850-851 (1986)). But the proposition that legal defenses based upon doctrines central to the courts’ structural independence can never be waived simply does not accord with our cases. Certainly one such doctrine consists of the “judicial Power” to disregard an unconstitutional statute, see Marbury, 1 Cranch, at 177; yet none would suggest that a litigant may never waive the defense that a statute is unconstitutional. See, e. g., G. D. Searle & Co. v. Cohn, 455 U. S. 404, 414 (1982). What may follow from our holding that the judicial power unalterably includes the power to render final judgments is not that waivers of res judicata are always impermissible, but rather that, as many Federal Courts of Appeals have held, waivers of res judicata need not always be accepted— that trial courts may in appropriate cases raise the res judicata bar on their own motion. See, e. g., Coleman v. Ramada Hotel Operating Co., 933 F. 2d 470, 475 (CA7 1991); In re Medomak Canning, 922 F. 2d 895, 904 (CA1 1990); Holloway Constr. Co. v. United States Dept. of Labor, 891 F. 2d 1211, 1212 (CA6 1989). Waiver subject to the control of the [232]*232courts themselves would obviously raise no issue of separation of powers, and would be precisely in accord with the language of the decision that the Solicitor General relies upon. We held in Schor that, although a litigant had consented to bring a state-law counterclaim before an Article I tribunal, 478 U. S., at 849, we would nonetheless choose to consider his Article III challenge, because “when these Article III limitations are at issue, notions of consent and waiver cannot be dispositive,” id., at 851 (emphasis added). See also Freytag v. Commissioner, 501 U. S. 868, 878-879 (1991) (finding a “rare cas[e] in which we should exercise our discretion” to hear a waived claim based on the Appointments Clause, Art. II, §2, cl. 2).6
Petitioners also rely on a miscellany of decisions upholding legislation that altered rights fixed by the final judgments of non-Article III courts, see, e. g., Sampeyreac v. United States, 7 Pet. 222, 288 (1833); Freeborn v. Smith, 2 Wall. 160 (1865), or administrative agencies, Paramino Lumber Co. v. Marshall, 309 U. S. 370 (1940), or that altered the prospective effect of injunctions entered by Article III courts, Wheeling & Belmont Bridge Co., 18 How., at 421. These cases distinguish themselves; nothing in our holding today calls them into question. Petitioners rely on general statements from some of these cases that legislative annulment of final judgments is not an exercise of judicial power. But even if it were our practice to decide cases by weight of prior dicta, we would find the many dicta that reject congressional [233]*233power to revise the judgments of Article III courts to be the more instructive authority. See supra, at 225-226.7
Finally, petitioners liken § 27A(b) to Federal Rule of Civil Procedure 60(b), which authorizes courts to relieve parties from a final judgment for grounds such as excusable neglect, newly discovered evidence, fraud, or “any other reason justifying relief . . . .” We see little resemblance. Rule 60(b), which authorizes discretionary judicial revision of judgments in the listed situations and in other “ ‘extraordinary circumstances,’” Liljeberg v. Health Services Acquisition Corp., 486 U. S. 847, 864 (1988), does not impose any legislative mandate to reopen upon the courts, but merely reflects and [234]*234confirms the courts’ own inherent and discretionary power, “firmly established in English practice long before the foundation of our Republic,” to set aside a judgment whose enforcement would work inequity. Hazel-Atlas Glass Co. v. Hartford-Empire Co., 322 U. S. 238, 244 (1944). Thus, Rule 60(b), and the tradition that it embodies, would be relevant refutation of a claim that reopening a final judgment is always a denial of property without due process; but they are irrelevant to the claim that legislative instruction to reopen impinges upon the independent constitutional authority of the courts.
The dissent promises to provide “[a] few contemporary examples” of statutes retroactively requiring final judgments to be reopened, “to demonstrate that [such statutes] are ordinary products of the exercise of legislative power.” Post, at 256. That promise is not kept. The relevant retroactivity, of course, consists not of the requirement that there be set aside a judgment that has been rendered prior to its being setting aside — for example, a statute passed today which says that all default judgments rendered in the future may be reopened within 90 days after their entry. In that sense, all requirements to reopen are “retroactive,” and the designation is superfluous. Nothing we say today precludes a law such as that. The finality that a court can pronounce is no more than what the law in existence at the time of judgment will permit it to pronounce. If the law then applicable says that the judgment may be reopened for certain reasons, that limitation is built into the judgment itself, and its finality is so conditioned. The present case, however, involves a judgment that Congress subjected to a reopening requirement which did not exist when the judgment was pronounced. The dissent provides not a single clear prior instance of such congressional áction.
The dissent cites, first, Rule 60(b), which it describes as a “familiar remedial measure.” Ibid. As we have just discussed, Rule 60(b) does not provide a new remedy at all, but [235]*235is simply the recitation of pre-existing judicial power. The same is true of another of the dissent’s examples, 28 U. S. C. § 2255, which provides federal prisoners a statutory motion to vacate a federal sentence. This procedure “‘restates, clarifies and simplifies the procedure in the nature of the ancient writ of error coram nobis.’” United States v. Hayman, 342 U. S. 205, 218 (1952) (quoting the 1948 Reviser’s Note to § 2255). It is meaningless to speak of these statutes as applying “retroactively,” since they simply codified judicial practice that pre-existed. Next, the dissent cites the provision of the Soldiers’ and Sailors’ Civil Relief Act of 1940, 54 Stat. 1178, 50 U. S. C. App. §520(4), which authorizes courts, upon application, to reopen judgments against members of the Armed Forces entered while they were on active duty. It could not be clearer, however, that this provision was not retroactive. It says: “If any judgment shall be rendered in any action or proceeding governed by this section against any person in military service during the period of such service . . . such judgment may ... be opened . . . .” (Emphasis added.)
The dissent also cites, post, at 258, a provision of the Handicapped Children’s Protection Act of 1986, 82 Stat. 901, 20 U. S. C. § 1415(e)(4)(B) (1988 ed. and Supp. V), which provided for the award of attorney’s fees under the Education for All Handicapped Children Act of 1975, 89 Stat. 773, 20 U. S. C. § 1411 et seq. (1988 ed. and Supp. V). This changed the law regarding attorney’s fees under the Education for All Handicapped Children Act, after our decision in Smith v. Robinson, 468 U. S. 992 (1984), found such fees to be unavailable. The provision of the Statutes at Large adopting this amendment to the United States Code specified, in effect, that it would apply not only to proceedings brought after its enactment, but also to proceedings pending at the time of, or brought after, the decision in Smith. See 100 Stat. 798. The amendment says nothing about reopening final judgments, and the retroactivity provision may well mean noth[236]*236ing more than that it applies not merely to new suits commenced after the date of its enactment, but also to previously filed (but not yet terminated) suits of the specified sort. This interpretation would be consistent with the only case the dissent cites, which involved a court-entered consent decree not yet fully executed. Counsel v. Dow, 849 P. 2d 731, 734, 738-739 (CA2 1988). Alternatively, the statute can perhaps be understood to create a new cause of action for attorney’s fees attributable to already concluded litigation. That would create no separation-of-powers problem, and would be consistent with this Court’s view that “[attorney's fee determinations ... are ‘collateral to the main cause of action’ and ‘uniquely separable from the cause of action to be proved at trial.’ ” Landgraf v. USI Film Products, 511 U. S., at 277 (quoting White v. New Hampshire Dept. of Employment Security, 455 U. S. 445, 451-452 (1982)).8
The dissent’s perception that retroactive reopening provisions are to be found all about us is perhaps attributable to its inversion of the statutory presumption regarding retroactivity. Thus, it asserts that Rule 60(b) must be retroactive, since “[n]ot a single word in its text suggests that it does not apply to judgments entered prior to its effective date.” [237]*237Post, at 256-257. This reverses the traditional rule, confirmed only last Term, that statutes do not apply retroactively unless Congress expressly states that they do. See Landgraf, supra, at 277-280. The dissent adds that “the traditional construction of remedial measures . . . supports] construing [Rule 60(b)] to apply to past as well as future judgments.” Post, at 257. But reliance on the vaguely remedial purpose of a statute to defeat the presumption against retroactivity was rejected in the companion cases of Landgraf, see 511 U. S., at 284-286, and n. 37, and Rivers v. Roadway Express, 511 U. S., at 309-313. Cf. Landgraf, supra, at 297 (Blackmun, J., dissenting) (“This presumption [against retroactive legislation] need not be applied to remedial legislation ...”) (citing Sampeyreac, 7 Pet., at 238).
The dissent sets forth a number of hypothetical horribles flowing from our assertedly “rigid holding” — for example, the inability to set aside a civil judgment that has become final during a period when a natural disaster prevented the timely filing of a certiorari petition. Post, at 262. That is horrible not because of our holding, but because the underlying statute itself enacts a “rigid” jurisdictional bar to entertaining untimely civil petitions. Congress could undoubtedly enact prospective legislation permitting, or indeed requiring, this Court to make equitable exceptions to an otherwise applicable rule of finality, just as district courts do pursuant to Rule 60(b). It is no indication whatever of the invalidity of the constitutional rule which we announce, that it produces unhappy consequences when a legislature lacks foresight, and acts belatedly to remedy a deficiency in the law. That is a routine result of constitutional rules. See, e. g., Collins v. Youngblood, 497 U. S. 37 (1990) (Ex Post Facto Clause precludes postoffense statutory extension of a criminal sentence); United States Trust Co. of N. Y. v. New Jersey, 431 U. S. 1 (1977) (Contract Clause prevents retroactive alteration of contract with state bondholders); Louisville Joint Stock Land Bank v. Radford, 295 U. S. 555, 589-[238]*238590, 601-602 (1935) (Takings Clause invalidates a bankruptcy law that abrogates a vested property interest). See also United States v. Security Industrial Bank, 459 U. S. 70, 78 (1982).
Finally, we may respond to the suggestion of the concurrence that this case should be decided more narrowly. The concurrence is willing to acknowledge only that “sometimes Congress lacks the power under Article I to reopen an otherwise closed court judgment,” post, at 240-241. In the present context, what it considers critical is that §27A(b) is “exclusively retroactive” and “appli[es] to a limited number of individuals.” Post, at 241. If Congress had only “provided] some of the assurances against ‘singling out’ that ordinary legislative activity normally provides — say, prospectivity and general applicability — we might have a different case.” Post, at 243.
This seems to us wrong in both fact and law. In point of fact, § 27A(b) does not “single out” any defendant for adverse treatment (or any plaintiff for favorable treatment). Rather, it identifies a class of actions (those filed pr e-Lampf, timely under applicable state law, but dismissed as time barred post-Lampf) which embraces many plaintiffs and defendants, the precise number and identities of whom we even now do not know. The concurrence’s contention that the number of covered defendants “is too small (compared with the number of similar, uncovered firms) to distinguish meaningfully the law before us from a similar law aimed at a single closed case,” post, at 244 (emphasis added), renders the concept of “singling out” meaningless.
More importantly, however, the concurrence’s point seems to us wrong in law. To be sure, the class of actions identified by § 27A(b) could have been more expansive (e. g., all actions that were or could have been filed pr e-Lampf) and the provision could have been written to have prospective as well as retroactive effect (e. g., “all post-Lampf dismissed actions, plus all future actions under Rule 10b-5, shall be timely if brought within 30 years of the injury”). But it escapes us [239]*239how this could in any way cause the statute to be any less an infringement upon the judicial power. The nub of that infringement consists not of the Legislature’s acting in a particularized and hence (according to the concurrence) nonlegislative fashion;9 but rather of the Legislature’s nullifying prior, authoritative judicial action. It makes no difference whatever to that separation-of-powers violation that it is in gross rather than particularized (e. g., “we hereby set aside all hitherto entered judicial orders”), or that it is not accompanied by an “almost” violation of the Bill of Attainder Clause, or an “almost” violation of any other constitutional provision.
Ultimately, the concurrence agrees with our judgment only “[bjecause the law before us embodies risks of the very sort that our Constitution’s ‘separation of powers’ prohibition seeks to avoid.” Post, at 246. But the doctrine of separation of powers is a structural safeguard rather than a remedy to be applied only when specific harm, or risk of specific harm, can be identified. In its major features (of which the conclusiveness of judicial judgments is assuredly one) it is a prophylactic device, establishing high walls and clear distinctions because low walls and vague distinctions will not be judicially defensible in the heat of interbranch conflict. It is interesting that the concurrence quotes twice, and cites without quotation a third time, the opinion of Justice Powell in [240]*240INS v. Chadha, 462 U. S., at 959. But Justice Powell wrote only for himself in that case. He alone expressed dismay that “[t]he Court’s decision . . . apparently will invalidate every use of the legislative veto,” and opined that “[t]he breadth of this holding gives one pause.” Ibid. It did not give pause to the six-Justice majority, which put an end to the long-simmering interbranch dispute that would otherwise have been indefinitely prolonged. We think legislated invalidation of judicial judgments deserves the same categorical treatment accorded by Chadha to congressional invalidation of executive action. The delphic alternative suggested by the concurrence (the setting aside of judgments is all right so long as Congress does not “impermissibly tr[y] to apply, as well as make, the law,” post, at 241) simply prolongs doubt and multiplies confrontation. Separation of powers, a distinctively American political doctrine, profits from the advice authored by a distinctively American poet: Good fences make good neighbors.
* * *
We know of no previous instance in which Congress has enacted retroactive legislation requiring an Article III court to set aside a final judgment, and for good reason. The Constitution’s separation of legislative and judicial powers denies it the authority to do so. Section 27A(b) is unconstitutional to the extent that it requires federal courts to reopen final judgments entered before its enactment. The judgment of the Court of Appeals is affirmed.
It is so ordered.