with whom The Chief Justice, Justice Scalia, and Justice Thomas join, dissenting.
California has enacted a retroactive extension of statutes of limitations for serious sexual offenses committed against minors. Cal. Penal Code Ann. § 803(g) (West Supp. 2003). The new period includes cases where the limitations period has expired before the effective date of the legislation. To invalidate the statute in the latter circumstance, the Court tries to force it into the second category of Calder v. Bull, 3 Dall. 386 (1798), which prohibits a retroactive law “ ‘that aggravates a crime, or makes it greater than it was, when committed.’” Ante, at 612 (quoting Calder, supra, at 390 (emphasis in original)). These words, in my view, do not permit the Court’s holding, but indeed foreclose it. A law which does not alter the definition of the crime but only revives prosecution does not make the crime “greater than it was, when committed.” Until today, a plea in bar has not been thought to form any part of the definition of the offense.
To overcome this principle, the Court invokes “a long line of authority holding that a law of this type violates the Ex Post Facto Clause.” Ante, at 621. The Court’s list of precedents, ante, at 617-619, is less persuasive than it may appear at a first glance. Of the 22 cases cited by the Court, only 4 had to decide whether a revival of expired prosecutions was constitutional. See Moore v. State, 43 N. J. L. 203, 216-217 (1881); United States v. Fraidin, 63 F. Supp. 271, 276 (Md. 1945); People v. Shedd, 702 P. 2d 267, 268 (Colo. 1985) (en banc) (per curiam); Commonwealth v. Rocheleau, 404 Mass. 129, 130-131, 533 N. E. 2d 1333, 1334 (1989), cited ante, at 617. These four cases — which are the only cases that are relevant — will be discussed in due course.
*634The case of State v. Sneed, 25 Tex. Supp. 66, 67 (1860), cited ante, at 617, is inapposite. There, the court avoided the issue by holding that the statute was not meant to apply retroactively. Interpreting the statute so as to avoid invalidation on constitutional grounds, Sneed did not pass on the merits. Even if the court addressed the merits, its cursory paragraph-long opinion, reproduced by the majority in its entirety, ante, at 629, contains no reference to Justice Chase’s classification, nor indeed any analysis whatsoever. This unreasoned opinion scarcely supports the majority’s novel interpretation of Calder’s second category.
In the remaining 17 eases, the question was not presented. As the Court itself concedes, eight of these cases considered only extensions of unexpired statutes of limitations, and upheld them. Ante, at 618. The Court does not mention that nine other cases have done so as well. See People ex rel. Reibman v. Warden, 242 App. Div. 282, 275 N. Y. S. 59 (1934); State v. Hodgson, 108 Wash. 2d 662, 740 P. 2d 848 (1987) (en banc); State v. Nunn, 244 Kan. 207, 768 P. 2d 268 (1989); State v. O’Neill, 118 Idaho 244, 796 P. 2d 121 (1990); State v. Schultzen, 522 N. W. 2d 833 (Iowa 1994); State v. Comeau, 142 N. H. 84, 697 A. 2d 497 (1997); State v. Hamel, 138 N. H. 392, 643 A. 2d 953 (1994); Santiago v. Commonwealth, 428 Mass. 39, 697 N. E. 2d 979 (1998), cited ante, at 617. Because these cases did not need to decide whether the Ex Post Facto Clause would bar the extension of expired limitations periods, the question did not receive the same amount of attention as if the courts were required to dispose of the issue.
The case law compiled by the Court is deficient, furthermore, at a more fundamental level. Our precedents hold that the reach of the Ex Post Facto Clause is strictly limited to the precise formulation of the Colder categories. We have made it clear that these categories provide “an exclusive definition of ex post facto laws,” Collins v. Youngblood, 497 U. S. 37, 42 (1990), and have admonished that it is *635“a mistake to stray beyond Calder’s four categories,” Carmell v. Texas, 529 U. S. 513, 539 (2000). Justice Chase himself stressed that the categories must be construed with caution to avoid any unnecessary extension: “I am under a necessity to give a construction, or explanation of the words, ‘ex post facto law,’ because they have not any certain meaning attached to them. But I will not go farther than I feel myself bound to do; and if I ever exercise the jurisdiction I will not decide any law to be void, but in a very clear case.” 3 Dall., at 395.
The Court seems to recognize these principles, ante, at 611-612, but then relies on cases which flatly contradict them. The opinion of the New Jersey’s Court of Errors and Appeals in Moore v. State, supra, on which the Court places special emphasis, see ante, at 613, 617, 628, 630, 632, expressly stated that a statute reviving an expired limitations period “is not covered by any of [Justice Chase’s] classes.” 43 N. J. L., at 216. The Moore court made a fleeting mention that the statute might fall within Chase’s fourth category, but immediately dismissed this line of inquiry. Instead, it proceeded to “[l]oo[k] away from his classification to what he states to have been the motive for and principle sustaining the edict.” Ibid. As Collins and Carmell explained, this expansive approach to the Ex Post Facto Clause is contrary to Calder’s admonition that its categories must be followed with care.
The majority’s lengthy defense of Moore’s legitimacy, ante, at 628, exposes the weaknesses both of that case and of the Court’s opinion. The majority argues Moore’s statement that the statute was not covered by Justice Chase’s categories referred only to the principal description of these categories, but not to the alternative one the Court now seeks to embrace. The view that a statute not covered by Justice Chase’s main formulations — the only formulations our cases have treated as authoritative — may still be ex post facto if it falls within his historical examples is a view no court until *636today has endorsed. The Moore court was no exception. When it held that the state statute was “not covered by any of [Justice Chase’s] classes,” Moore made clear it was looking beyond the language of the Colder categories: “Judge Chase did not consider his classes as exhaustive,” and so “a statute substantially imposing punishment for a previous act which, without the statute, would not be so punishable, is an ex post facto law, although it may not be included in the letter of Judge Chase’s rules.” 43 N. J. L., at 216, 220. The point was further emphasized by the separate opinion of Chancellor Runyon, a member of the one-judge Moore majority that invalidated the law as ex post facto: “[W]here the enactment, in whatever guise legislative ingenuity or subtlety may present it, inflicts the substantial injury, and does the essential wrong which the constitution sought to guard against, a true interpretation will hold it to be within the prohibition.” Id., at 226. The references to “substantia[l] imposition of] punishment” and “substantial injury” are reminiscent of the references to “substantial protections” and “substantial personal rights” used to enlarge the scope of the Ex Post Facto Clause and disapproved of in Collins. 497 U. S., at 46. By endorsing Moore, the majority seeks to resurrect this rejected reasoning here.
The other precedents the Court invokes — both the cases where extension of expired statutes of limitations was at issue and the cases which merely opined on the question in dicta — have the same flaw. The misconception causing it arises from Judge Learned Hand’s dictum, mentioned while holding that an extension of an unexpired statute of limitations is not ex post facto, that if the statute had expired there would be a violation. Falter v. United States, 23 F. 2d 420, 425 (CA2 1928). Judge Hand based this distinction on a citation of the faulty decision in Moore and on his belief that whether an extension of a limitations period is ex post facto “turns upon how much violence is done to our instinctive feelings of justice and fair play.” Falter, supra, at 425-426. *637The Court’s opinion is premised on the same approach. It relies on Judge Hand for the proposition that an extension of expired limitations periods “ ‘seems to most of us unfair and dishonest.’” Ante, at 611 (quoting Falter, supra, at 426). In previous cases, however, the Court has explained that this conception of our ex post facto jurisprudence is incorrect: “[W]hile the principle of unfairness helps explain and shape the Clause’s scope, it is not a doctrine unto itself, invalidating laws under the Ex Post Facto Clause by its own force.” Carmell, supra, at 533, n. 23 (citing W. S. Kirkpatrick & Co. v. Environmental Tectonics Corp., Int’l, 493 U. S. 400, 409 (1990)).
It was the unsupported Hand observation that formed the rationale applied by many of the cases the Court cites, including all the post-Moore cases where expired limitations periods were at issue. See Fraidin, 63 F. Supp., at 276 (relying on Falter and containing no discussion of the Calder categories); Shedd, 702 P. 2d, at 268 (same); Hodgson, 108 Wash. 2d, at 667-668, 740 P. 2d, at 851 (relying on, and quoting from, Falter); Rocheleau, 404 Mass., at 130, 533 N. E. 2d, at 1334 (containing no Calder analysis but relying instead on its earlier decision in Commonwealth v. Bargeron, 402 Mass. 589, 524 N. E. 2d 829 (1988), which in turn was based on Falter); O'Neill, 118 Idaho, at 246, 796 P. 2d, at 123 (citing Falter and supplying no analysis of its own); State v. Hirsch, 245 Neb. 31, 39, 511 N. W. 2d 69, 76 (1994) (relying on Falter); Hamel, 138 N. H., at 395, 643 A. 2d, at 955 (same). Since these cases applied the methodology our Court has disavowed, they provide the majority with scant support. None of them even discussed the issue in terms of Calder's second category, much less construed that category in the manner today’s decision improperly proposes. The flaw of these cases is not, as the majority argues, that they are “not perfectly consistent with modern conceptions of how legal analysis should proceed,” ante, at 630; the flaw is that their method of analysis is foreclosed by this Court’s precedents.
*638The majority turns for help to a roster of commentators who concluded that revival of expired statutes of limitations is precluded by the ex post facto guarantee. See ante, at 619-620. Some of the commentators applied the same expansive approach we have declared impermissible in Collins and Carmell. Henry Black, on whose work the Court relies the most, see ante, at 613, 615, 619, openly acknowledged that the revival of expired statutes of limitations is not covered by any of the Colder categories. See Constitutional Prohibitions Against Legislation Impairing the Obligation of Contracts, and Against Retroactive and Ex Post Facto Laws § 227, p. 291 (1887). Black, moreover, relied on the example of the civil statutes of limitations, which he believed could not be revived. Id., §235, at 296-297. The Court’s later case law has rendered this interpretation questionable. See, e. g., Chase Securities Corp. v. Donaldson, 325 U. S. 304, 314-316 (1945). Other commentators relied, often with no analysis, on the Moore and Falter line of cases, which were plagued by methodological infirmities since discovered. See authorities cited ante, at 619. None of these scholars explained their conclusion by reference to Colder’s second category.
There are scholars who have considered with care the meaning of that category; and they reached the conclusion stated in this dissent, not the conclusion embraced by the majority. In his treatise on retroactive legislation, William Wade defined the category as covering the law “which undertakes to aggravate a past offence, and make it greater than when committed, endeavors to bring it under some description of transgression against which heavier penalties or more severe punishments have been denounced: as, changing the character of an act which, when committed, was a misdemeanor, to a crime; or, declaring a previously committed of-fence, of one of the classes graduated, and designated by the number of its degree, to be of a higher degree than it was when committed.” Operation and Construction of Retroac*639tive Laws §273, pp. 317-318 (1880). Joel Prentiss Bishop’s work on statutory crimes concluded that a law reviving expired prosecution “is not within any of the recognized legal definitions of an ex post facto law.” Commentaries on the Law of Statutory Crimes §266, p. 294 (rev. 3d ed. 1901). The author’s explanation is an apt criticism of the Court’s opinion: “The punishment which it renders possible, by forbidding the defense of lapse of time, is exactly what the law provided when ‘the fact’ transpired. No bending of language, no supplying of implied meanings, can, in natural reason, work out the contrary conclusion. . . . The running of the old statute had taken from the courts the right to proceed against the offender, leaving the violated law without its former remedy; but it had not obliterated the fact that the law forbade the act when it was done, or removed from the doer’s mind his original consciousness of guilt.” Id., § 266, at 294-295. In reaching his conclusion, Bishop considered, and rejected, the argument put forth by the Moore majority. Commentaries on the Law of Statutory Crimes, supra, § 266, at 295, and n. 5. This rejection does not, as the majority believes, undermine Bishop’s conclusion, see ante, at 630; given Moore’s infirmities, it strengthens the validity of his interpretation.
This definition of Calder's second category is necessary for consistency with our accepted understanding of categories one and three. The first concerns laws declaring innocent acts to be a crime; the third prohibits retroactive increases in punishment. 3 Dall., at 390. The first three categories guard against the common problem of retroactive redefinition of conduct by criminalizing it (category one), enhancing its criminal character (category two), or increasing the applicable punishment (category three). The link between these categories was noted by Justice Paterson in Calder itself: “The enhancement of a crime, or penalty, seems to come within the same mischief as the creation of a crime or pen*640alty; and therefore they may be classed together.” Id., at 397.
The point is well illustrated in Beazell v. Ohio, 269 U. S. 167 (1925), whose formulation of the Calder categories we later described as “faithful to our best knowledge of the original understanding of the Ex Post Facto Clause.” Collins, 497 U. S., at 43. Beazell involved a retroactively applied law providing for joint trials for most felonies, with separate trials allowed only when requested by one of the defendants or the prosecutor, and only with the leave of the court. 269 U. S., at 168-169. The prior law had provided for separate trials whenever a defendant so requested. Id., at 168. Reviewing an ex post facto challenge to the new law, the Court noted that the first three Calder categories address “the criminal quality attributable to an act.” 269 U. S., at 170. Applying this definition, the Court held the state statute did not violate the Ex Post Facto Clause because “[i]t does not deprive [the defendant] of any defense previously available, nor affect the criminal quality of the act charged. Nor does it change the legal definition of the offense or the punishment to be meted out.” Ibid. In other words, the Ohio statute fell into none of the first three Calder categories. The second category, as the Beazell Court understood it, covered those retroactive statutes which “affect the criminal quality of the act charged [by] changing] the legal definition of the offense.” 269 U. S., at 170. The California statute challenged by petitioner changes only the timespan within which the action against him may be filed; it does not alter the criminal quality assigned to the offense.
The Court’s opinion renders the second Calder category unlimited and the surrounding categories redundant. A law which violates the first Calder category would also violate the Court’s conception of category two, because such a law would “inflic[t] punishments, where the party was not, by law, liable to any punishment.” Ante, at 612 (emphasis deleted and internal quotation marks omitted). The majority *641attempts to eliminate this redundancy by limiting its definition to instances where the conduct was criminal, yet if Justice Chase’s alternative description of the second category is supposed to be definitive of its scope, ante, at 611, it would seem to strike broader than the Court’s limiting construction. Similarly, a retroactive law increasing punishment in violation of the third category would also constitute an “innovation” for which, prior to the passage of the new law, the offender was not liable, ante, at 612, and so be prohibited under the Court’s unbounded interpretation of category two. The Court’s new definition not only distorts the original meaning of the second Calder category, but also threatens the coherence of the overall ex post facto scheme.
Realizing the inconsistency, the majority scarcely refers to the authoritative language Justice Chase used to describe the second category. Instead, the Court relies on what it terms Justice Chase’s alternative description of that category, which speaks about laws which “ ‘inflict[ed] punishments, where the party was not, by law, liable to any punishment.’” Ante, at 612 (emphasis deleted) (quoting Calder, 3 Dall., at 389). These words are not, strictly speaking, a description of the second category itself; they are a description of the category’s historical origins. Justice Chase used them to refer to certain laws passed by the British Parliament which led the Founders to adopt the Ex Post Facto Clause; he did not intend them as a definitive description of the laws prohibited by that constitutional provision. Ibid. This description of a category’s origins may, of course, shed light on the meaning of Justice Chase’s principal formulation, which was meant to be definitive. The Court, however, uses Chase’s alternative description as the independent operative definition of that category. None of our precedents, until today, based their holding on the language of Justice Chase’s alternative description, certainly not in situations when the statute under review would not fit within the principal formulation.
*642The Court, in any event, misunderstands the alternative description. As our precedents have instructed, this description must be viewed in the context of the history of the British parliamentary enactments to which Justice Chase referred. Ante, at 614; cf. Carmell, 529 U. S., at 526-530 (examining the historical circumstances of the case of Sir John Fenwick, cited by Justice Chase as an example of the fourth ex post facto category, in order “[t]o better understand the type of law that falls within that category”). With respect to the second category, Justice Chase provided two examples: the banishments of Lord Clarendon in 1667 and of Bishop Francis Atterbury in 1723. Calder, supra, at 389, and n. ‡ (citing 19 Car. II, c. 10; 9 Geo. I, c. 17). A consideration of both historical episodes confirms that Calder’s second category concerns only laws which change the nature of an offense to make it greater than it was at the time of commission, thereby subjecting the offender to increased punishment.
Justice Chase and, it can be presumed, the Founders were familiar with the parliamentary proceedings leading to the banishment of the Earl of Clarendon. Clarendon, former Lord Chancellor and principal advisor to Charles II, was impeached by the House of Commons on charges of treason. Edward Earl of Clarendon’s Trial, 6 How. St. Tr. 292, 330-334, 350 (1667) (hereinafter Clarendon’s Trial)', G. Miller, Edward Hyde, Earl of Clarendon 20-21 (1983). The House of Lords, however, refused to commit Clarendon to trial, finding the allegations not cognizable as treason under the law. Clarendon’s Trial 358, 367. With the two Houses deadlocked, Clarendon left the country, an exit wise for his safety, perhaps, but not for his cause. For upon his departure the impeachment was abandoned yet Parliament agreed on a bill banishing Clarendon for treason and imposing an extensive range of civil disabilities. Id., at 374, 385, 390-391.
*643The principal objection raised against the impeachment charges was that they did not, under the law of the time, constitute treason. Id., at 342-346, 348-349, 350, 356-360, 367-372. The objection was not, it must be noted, that the charges were premised on innocent conduct. (If that were the nature of the objection, Justice Chase would have used the case to illustrate his first category, rather than his second one.) In fact, the impeachment explicitly alleged that Clarendon violated the law. See id., at 330-333. The objection made by Chase and by later legal scholars was that by the act of banishment the House sought to elevate criminal behavior of lower magnitude to the level of treason, thereby redefining what constitutes a treasonous offense. Even if Parliament assumed, on the basis of Clarendon’s flight, that the allegations were true, see id., at 389-390, that constructive admission did not alter the fact that, under the laws of the time, the allegations could not support a charge of treason. By enacting the bill, Parliament declared these allegations sufficient to constitute treason. Some parliamentary colloquy suggested, moreover, that Clarendon was being punished for his flight, rather than for offenses alleged. See id., at 389 (“[I]t is plain, if you proceed upon this bill, you go not upon your impeachment, but because he is fled from the justice of the land”). A flight from justice was not considered an offense so severe as to warrant banishment, “the highest punishment next to death.” Id., at 386. If the offense of flight was enhanced because of the prior offenses, then it was an increase in the gravity of the crime after its commission. Either way, the legislation increased the gravity of Clarendon’s offense.
The bill passed against Clarendon accomplished what English common-law scholar Richard Wooddeson described as the danger against which the second ex post facto category was designed to guard. The bill “ma[de] some innovation, or creat[ed] some forfeiture or disability, not incurred in the ordinary course of law.” 2 A Systematical View of the Laws *644of England 638 (1792) (hereinafter Wooddeson). It was Wooddeson’s interpretation of the English common law that Justice Chase relied upon. See Colder, 3 Dall., at 391; Carmell, supra, at 522-523, and n. 10; ante, at 614. The Court argues that the innovation deplored by Wooddeson was the imposition of a sanction (banishment) which, under settled law, was the prerogative of Parliament, not of the courts. Ibid. That may be so, but it cannot help the Court because this is not what California has done. Section 803(g) did not impose any punishment not otherwise contained in the California Penal Code. It did what legislatures have done throughout history: It specified when the criminal justice system may prosecute certain crimes. The majority tries to explain away this distinction as “not determinative,” ibid., but it makes all the difference. By imposing on a particular offender a punishment not prescribed by the existing legal norms a legislature signals its judgment that the gravity of the offense warrants its special intervention. In contrast, by prescribing general rules for the adjudication of offenses the legislature leaves the determination of the offender’s culpability entirely to the courts.
The majority’s explanation of the English precedents, in all events, is not the most logical one. Justice Chase’s alternative description covered enactments which “inflicted punishments, where the party was not, by law, liable to any punishment.” Colder, supra, at 389. Though only a parliamentary Act could subject an individual to banishment in 17th-century England, Parliament’s power to pass such Acts was unquestioned. See 11 W. Holdsworth, A History of English Law 569 (1938). A sanction of banishment was acknowledged as a punishment provided for by the existing laws, both at the time of Clarendon’s trial and afterwards. See, e.g., Craies, Compulsion of Subjects to Leave the Realm, 6 L. Q. Rev. 388, 392 (1890) (“[Bjanishment, perpetual or temporary, was well known to the common law”); An Act for Punishment of Rogues, 39 Eliz. 1, c. 4, s. 4 (1597) (permit*645ting banishment of dangerous rogues); the Roman Catholic Relief Act, 10 Geo. 4, c. 7, s. 28 (1829) (providing for the banishment of Jesuits). By law, then, a charge of high treason would have made Clarendon liable to banishment, which is inconsistent with Justice Chase’s formulation.
To explain away the inconsistency, the Court redefines the words “by law” to refer only to punishments “not otherwise available ‘in the ordinary course of law.’” Ante, at 614 (quoting 2 Wooddeson 638). As already explained, it was an accepted procedure in 17th-century England for Parliament to pass laws imposing banishment.
The majority must mean, then, that banishment was not available through the courts. At the time of Clarendon’s trial, however, British courts were empowered to adjudicate treason and to punish it with death. 1 M. Hale, Pleas of the Crown *348-*351; see also 2 Jowitt’s Dictionary of English Law 1799-1800 (2d ed. 1977). If the charges against Clarendon accurately alleged treason, he was eligible, through ordinary judicial proceedings, to receive capital punishment, which was obviously a sanction more severe than banishment. For the majority’s historical explanation to work, Justice Chase’s alternative description of the second category would have to prohibit laws which inflicted a punishment where the party was not, through normal judicial proceedings, liable to that precise punishment but was liable to a greater one. This formulation can hardly be reconciled with the words Justice Chase used, much less with his principal formulation of the second category. A legislature does not make an individual’s crime “greater than it was, when committed,” Calder, supra, at 390, by assigning a punishment less severe than the one available through the courts.
If Justice Chase’s reference to Clarendon’s trial is to have explanatory power, one must look for an alternative interpretation. What was repulsive to Chase and Wooddeson in Clarendon’s trial was not the imposition of banishment as such, but that the sanction was outside the limits of what *646Clarendon’s offense merited under the law established at the time of its commission, and was instead premised on Parliament’s exaggeration of the gravity of the offense. Viewed this way, the Clarendon example lends no support to the majority’s position, but instead undercuts it.
It must be acknowledged that, as the majority points out, a number of historians have treated one of the charges levied against Clarendon, that of betraying the King’s secrets to the enemy, as impeachable treason. Ante, at 622-623. The historical judgment, however, is not as uniform as the Court makes it seem. See 7 E. Foss, Judges of England 130 (1864) (“No one can read the articles [against Clarendon] without seeing the. weakness and frivolity of the allegations, none of them, even if true, amounting to treason”); R. Berger, Impeachment: The Constitutional Problems 45-46 (1974) (explaining the articles of impeachment against Clarendon as based on the Parliament’s power to declare certain nontrea-sonous offenses to be treason).
Historians are in agreement, though, that the Commons could not substantiate the charge of betraying secrets to the enemy. 2 H. Hallam, Constitutional History of England: From the Accession of Henry VII to the Death of George II 367, 373 (rev. ed. 1881); Roberts, The Impeachment of the Earl of Clarendon, 13 Camb. Hist. J. 13-14 (1957); Roberts, The Law of Impeachment in Stuart England, 84 Yale L. J. 1419, 1426 (1975); Berger, supra, at 45, n. 193. It is due to this absence of evidence that the Commons refused to produce particulars of the treason charge against Clarendon, insisting instead the Lords trust their word that the underlying conduct was treasonous. Although the technical grounds for the Lords’ objection to this charge was the lack of specificity, the objection can also be viewed as reflecting a belief that the Commons were attempting to aggravate Clarendon’s offenses by labeling them as treason absent any justification. As Henry Hallam has explained in his respected study of the English constitutional history, “if the *647house of lords shall be of opinion, either by consulting the judges or otherwise, that no treason is specially alleged, they should, notwithstanding any technical words, treat the of-fence as a misdemeanor.” 2 Hallam, supra, at 413. Justice Chase could have viewed the betrayal of secrets charge in a similar way, as a subterfuge through which the Commons were trying to elevate Clarendon’s offenses to the level of treason.
The proposed interpretation of Clarendon’s example is reinforced by considering the proceedings against Bishop Francis Atterbury, who, in the midst of hysteria over both real and supposed Jacobite plots, was accused of conspiracy to depose George I. The evidence against Atterbury was meager, and supporters of the Crown, fearing that neither the common-law courts nor even the House of Lords would convict, introduced a bill of banishment. G. Bennett, Tory Crisis in Church and State, 1688-1730, pp. 258-265 (1975); Bishop Atterbury’s Trial, 16 How. St. Tr. 323,640 (1723) (reprint 2000) (hereinafter Atterbury’s Trial). The bill declared Atterbury a traitor, and subjected him to a range of punishments not previously imposed, including exile and civil death. Id., at 644-646; Bennett, supra, at 265. The Duke of Wharton, who registered the lengthiest dissent, commented that “this Bill seems as irregular in the punishments it inflicts, as it is in its foundation, and carries with it an unnatural degree of hardship.” Atterbury’s Trial 691. The only bill of comparable harshness was the Act banishing Clarendon. Those sanctions were more mild, id., at 691-692, but, as we have seen, just as violative of the rule against penalties imposed after the fact. As in the case of Clarendon, Parliament adjudged Atterbury’s offense to be so grave as to merit a singularly severe punishment. The bill designed vindictive forfeitures and disabilities not imposed in the ordinary course of law.
The Atterbury case illustrates again the close relationship between the second and the third Calder categories. See *648supra, at 639-640 (quoting Calder, 3 Dall., at 397 (Paterson, J.)). As already explained, supra, at 640-641, the Court’s misconstruction of Justice Chase’s historical examples takes the second category out of this logical continuum. Contrary to the majority’s belief, ante, at 625, an interpretation which highlights the link between these two categories is more faithful to the original understanding. Richard Wooddeson, the Court’s preferred commentator, discussed these two categories together, noting that both “principally affect the punishment.” 2 Wooddeson 638-640; see also id., at 624.
Atterbury’s trial also illustrates why the majority’s interpretation of the historical examples as premised on the courts’ inability to impose banishment is untenable. See supra, at 645-646. Had Atterbury been convicted of treason through the courts, he would have been subject to capital punishment. Parliament’s decision to prosecute Atterbury may have been driven by fear of backlash provoked by a death sentence, for Atterbury enjoyed considerable popularity and sympathy in some circles. See Bennett, supra, at 259. Wooddeson speculated, in an observation in tension with the majority’s interpretation, that Atterbury’s sentence may have been motivated by a desire “of mitigating punishment.” 2 Wooddeson 639. The mitigation, of course, was in comparison to the possible death verdict, not, as already explained, in comparison to the ordinary noncapital punishment Atterbury could have received.
Clarendon’s and Atterbury’s trials show why Stogner’s case does not belong in Calder’s second ex post facto category. The California Legislature did not change retroactively the description of Stogner’s alleged offense so as to subject him to an unprecedented and particularly severe punishment. The offense is described in the same terms as before the passage of § 803(g); the punishment remains the same. The character of the offense is therefore unchanged; it is perceived by the criminal justice system in the same way as before, and punished with the same force. The only *649change is that Stogner may now be prosecuted, whereas prior to the statute the prosecution could not have taken place. These illustrative examples, then, suggest the second Calder category encompasses only the laws which, to the detriment of the defendant, change the character of the offense to make it greater than it was at the time of commission.
The majority seems to suggest that retroactive extension of expired limitations periods is “‘arbitrary and potentially vindictive legislation,’” ante, at 611 (quoting Weaver v. Graham, 450 U. S. 24, 29, and n. 10 (1981)), but does not attempt to support this accusation. And it could not do so. The California statute can be explained as motivated by legitimate concerns about the continuing suffering endured by the victims of childhood abuse.
The California Legislature noted that “young victims often delay reporting sexual abuse because they are easily manipulated by offenders in positions of authority and trust, and because children have difficulty remembering the crime or facing the trauma it can cause.” People v. Frazer, 21 Cal. 4th 737, 744, 982 P. 2d 180, 183-184 (1999). The concern is amply supported by empirical studies. See, e. g., Summit, Abuse of the Child Sexual Abuse Accommodation Syndrome, in 1 J. of Child Sexual Abuse 153, 156-163 (1992); Lyon, Scientific Support for Expert Testimony on Child Sexual Abuse Accommodation, in Critical Issues in Child Sexual Abuse 107, 114-120 (J. Conte ed. 2002).
The problem the legislature sought to address is illustrated well by this case. Petitioner’s older daughter testified she did not report the abuse because she was afraid of her father and did not believe anyone would help her. After she left petitioner’s home, she tried to forget the abuse. Petitioner’s younger daughter did not report the abuse because she was scared. He tried to convince her it was a normal way of life. Even after she moved out of petitioner’s house, she was afraid to speak for fear she would not be *650believed. She tried to pretend she had a normal childhood. It was only her realization that the father continued to abuse other children in the family that led her to disclose the abuse, in order to protect them.
The Court tries to counter by saying the California statute is “ ‘unfair and dishonest”’ because it violated the State’s initial assurance to the offender that “ ‘he has become safe from its pursuit’” and deprived him of “the ‘fair warning.’” Ante, at 611 (quoting Falter v. United States, 23 F. 2d, at 426; Weaver, supra, at 28). The fallacy of this rationale is apparent when we recall that the Court is careful to leave in place the uniform decisions by state and federal courts to uphold retroactive extension of unexpired statutes of limitations against an ex post facto challenge. Ante, at 613.
There are two rationales to explain the proposed dichotomy between unexpired and expired statutes, and neither works. The first rationale must be the assumption that if an expired statute is extended, the crime becomes more serious, thereby violating category two; but if an unexpired statute is extended, the crime does not increase in seriousness. There is no basis in logic, in our cases, or in the legal literature to support this distinction. Both extensions signal, with equal force, the policy to prosecute offenders.
This leaves the second rationale, which must be that an extension of the expired statute destroys a reliance interest. We should consider whether it is warranted to presume that criminals keep calendars so they can mark the day to discard their records or to place a gloating phone call to the victim. The first expectation is minor and likely imaginary; the second is not, but there is no conceivable reason the law should honor it. And either expectation assumes, of course, the very result the Court reaches; for if the law were otherwise, there would be no legitimate expectation. The reliance exists, if at all, because of the circular reason that the Court today says so; it does not exist as part of our traditions or social understanding.
*651In contrast to the designation of the crime, which carries a certain measure of social opprobrium and presupposes a certain punishment, the statute of limitations has little or no deterrent effect. See Note, Retroactive Application of Legislatively Enlarged Statutes of Limitations for Child Abuse: Time’s No Bar to Revival, 22 Ind. L. Rev. 989, 1014 (1989) (“The statute of limitations has no measurable impact on allegedly criminal behavior, neither encouraging nor deterring such conduct”); Note, Ex Post Facto Limitations on Legislative Power, 73 Mich. L. Rev. 1491, 1513 (1975) (“[W]hile many defendants rely on substantive definitions of proscribed conduct, few rely on many of the numerous laws regulating the enforcement processes”). The Court does not claim a sex offender would desist if he knew he would be liable to prosecution when his offenses were disclosed.
The law’s approach to the analogous problem of reliance by wrongdoers in the civil sphere is instructive. We have held that expired statutes of limitations can be repealed to revive a civil action. See, e. g., Chase Securities Corp., 325 U. S., at 314; Plant v. Sprendthrift Farm, Inc., 514 U. S. 211, 229 (1995). These holdings were made in the areas of contracts and investments where reliance does exist and does matter. We allow the civil wrong to be vindicated nonetheless. If we do so in the civil sphere where reliance is real, we should do so in the criminal sphere where it is, for the most part, a fictional construct.
When a child molester commits his offense, he is . well aware the harm will plague the victim for a lifetime. See Briere & Runtz, Post Sexual Abuse Trauma: Data and Implications for Clinical Practice, 2 J. of Interpersonal Violence 367, 374-376 (1987); 1 J. Myers, Evidence in Child Abuse and Neglect Cases §4.2, pp. 221-223 (2d ed. 1992); Browne & Finkelhor, Initial and Long-Term Effects: A Review of the Research, in A Sourcebook on Child Sexual Abuse 143, 150-164 (D. Finkelhor et al. eds. 1986). The victims whose interests § 803(g) takes into consideration have been subjected to *652sexual abuse within the confines of their own homes and by people they trusted and relied upon for protection. A familial figure of authority can use a confidential relation to conceal a crime. The violation of this trust inflicts deep and lasting hurt. Its only poor remedy is that the law will show its compassion and concern when the victim at last can find the strength, and know the necessity, to come forward. When the criminal has taken distinct advantage of the tender years and perilous position of a fearful victim, it is the victim’s lasting hurt, not the perpetrator’s fictional reliance, that the law should count the higher. The victims whose cause is now before the Court have at last overcome shame and the desire to repress these painful memories. They have reported the crimes so that the violators are brought to justice and harm to others is prevented. The Court now tells the victims their decision to come forward is in vain.
The gravity of the crime was known, and is being measured, by its wrongfulness when committed. It is a common policy for States to suspend statutes of limitations for civil harms against minors, in order to “protec[t] minors during the period when they are unable to protect themselves.” 2 C. Corman, Limitation of Actions § 10.2.1, p. 104 (1991). Some States toll the limitations periods for minors even where a guardian is appointed, see id., at 105-106, and even when the tolling conflicts with statutes of repose, id., at 108. The difference between suspension and reactivation is so slight that it is fictional for the Court to say, in the given context, the new policy somehow alters the magnitude of the crime. The wrong was made clear by the law at the time of the crime’s commission. The criminal actor knew it, even reveled in it. It is the commission of the then-unlawful act that the State now seeks to punish. The gravity of the crime is left unchanged by altering a statute of limitations of which the actor was likely not at all aware.
*653The California statute does not fit any of the remaining Calder categories: It does not criminalize conduct which was innocent when done; it allows the prosecutor to seek the same punishment as the law authorized at the time the offense was committed and no more; and it does not alter the government’s burden to establish the elements of the crime. Any concern about stale evidence can be addressed by the judge and the jury, and by the requirement of proof beyond reasonable doubt. Section 803(g), moreover, contains an additional safeguard: It conditions prosecution on a presentation of independent evidence that corroborates the victim’s allegations by clear and convincing evidence. Cal. Penal Code Ann. §§ 803(g)(1), (2)(B) (West Supp. 2003). These protections, as well as the general protection against oppressive prosecutions offered by the Due Process Clause, should assuage the majority’s fear, ante, at 631, that the statute will have California overrun by vindictive prosecutions resting on unreliable recovered memories. See United States v. Lovasco, 431 U. S. 783, 789 (1977).
The statute does not violate petitioner’s rights under the Due Process Clause. We have held, in the civil context, that expired statutes of limitations do not implicate fundamental rights under the Clause. See, e. g., Chase Securities Corp., supra, at 314. For reasons already explained, see supra, at 652, there is no reason to reach a different conclusion here.
The Court’s stretching of Calder's second category contradicts the historical understanding of that category, departs from established precedent, and misapprehends the purposes of the Ex Post Facto Clause. The Court also disregards the interests of those victims of child abuse who have found the courage to face their abusers and bring them to justice. The Court’s opinion harms not only our ex post facto jurisprudence but also these and future victims of child abuse, and so compels my respectful dissent.