with whom Justice Scalia, •Justice Kennedy, and Justice Thomas join, dissenting.
The Court in this case conducts a truncated inquiry into a question of congressional intent, and, I believe, reaches the wrong result. The Court begins, uncontroversially enough, by observing that application of the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) to pending cases depends upon congressional intent, and that our inquiry into that intent should rely upon the “normal rules” of statutory construction. Ante, at 326. The Court then proceeds, however, to disregard all of our retroactivity case law — which it rather oddly disparages as manifestations of “Landgrafs default rule,” ibid. — in favor of a permissible, but by no means controlling, negative inference that it draws from the statutory text. I would instead interpret the AEDPA in light of the whole of our longstanding retroactivity jurisprudence, and accordingly find that the. amended 28 U. S. C. § 2254(d) (1994 ed., Supp. II) applies to pending cases.
The first question we must ask is whether Congress has expressly resolved whether the provision in question applies to pending cases. Landgraf v. USI Film Products, 511 U. S. 244, 280 (1994). Here, the answer is plainly no. The AEDPA does not clearly state, one way or the other, whether chapter 153 applies to pending cases. Given congressional silence, we must still interpret that statute, and that interpretation is in turn guided by the retroactivity principles we have developed over the years. The Court relies on one canon of statutory interpretation, expressio unius est exclusio alterius, to the exclusion of all others.
The Court’s opinion rests almost entirely on the negative inference that can be drawn from the fact that Congress expressly made chapter 154, pertaining to capital cases, ap*338plicable to pending cases, but did not make the same express provision in regards to chapter 153. That inference, however, is by no means necessary, nor is it even clearly the best inference possible. Certainly, Congress might have intended that omission to signal its intent that chapter 153 not apply to pending cases. But there are other, equally plausible, alternatives.
First, because chapter 154’s applicability is conditioned upon antecedent events — namely, a State’s establishing qualifying capital habeas representation procedures — Congress could have perceived a greater likelihood that, absent express provision otherwise, courts would fail to apply that chapter’s provisions to pending capital cases. Second, because of the characteristically extended pendency of collateral attacks on capital convictions,1 and because of Congress’ concern with the perceived acquiescence in capital defendants’ dilatory tactics by some federal courts (as evidenced by chapter 154’s strict time limits for adjudication of capital cases and, indeed, by the very title of the statute, the “Anti-terrorism and Effective Death Penalty Act of 1996”), Congress could very well have desired to speak with exacting clarity as to the applicability of the AEDPA to pending capital cases. Or third, Congress, while intending the AEDPA definitely to apply to pending capital cases, could have been uncertain or in disagreement as to which of the many portions of chapter 153 should or should not apply to pending cases. Congress could simply have assumed that the courts would sort out such questions, using our ordinary retroactivity presumptions.
None of these competing inferences is clearly superior to the others. The Court rejects the first, ante, at 330-332, as an “implausible” solution to an “unlikely” ambiguity. But *339the solution is not nearly as implausible as the Court’s contention that, in order to show that it wished chapter 153 not to apply to pending cases, Congress chose to make chapter 151p expressly applicable to such cases. If Congress wanted to make chapter 153 inapplicable to pending cases, the simplest.way to do so would be to say so. But, if Congress was instead concerned that courts would interpret chapter 154, because of its contingent nature, as not applying to pending cases, the most direct way to solve that concern would be the solution it adopted: expressly stating that chapter 154 did indeed apply to pending cases.
The Court finds additional support for its inference in the new 28 U. S. C. § 2264(b) (1994 ed., Supp. II), which it believes “tends to confirm,” ante, at 332, its analysis. Section 2264 is part of chapter 154 and forbids (subject to narrow exceptions) federal district courts to consider claims raised by state capital defendants unless those claims were first raised and decided on the merits in state court. Section 2264(b) provides, “[f]ollowing review subject to subsections (a), (d), and (e) of section 2254 [contained within chapter 153], the court shall rule on the claims properly before it.” This section, I believe, is irrelevant to the question before us.
The Court’s somewhat tortured interpretation of this section, as a backhanded way of making §§ 2254(a), (d), and (e) (but not the rest of chapter 153) apply to pending cases, is not convincing. For one thing, § 2264(b) is not phrased at all as a timing provision; rather than containing temporal language applying select sections to pending cases, § 2264(b) speaks in present tense, about how review should be conducted under chapter 154. Even more tellingly, as the Court implicitly concedes when it blandly describes this provision as a “loose end,” ante, at 336, the AEDPA did not alter § 2254(a), and so there is no need for an express provision making it applicable to pending cases.
Chapter 154 establishes special procedures for capital prisoners. Section 2264(b), by its terms, makes clear that *340§§ 2254(a), (d), and (e) apply to chapter 154 proceedings. That clarification makes sense in light of § 2264(a), which replaces the exhaustion requirement of §§ 2254(b) and (c) with a requirement that federal courts consider (subject to narrow exceptions) only those claims “raised and decided on the merits in the State courts.” Without that clarification, doubt might exist as to whether the rest of § 2254 still applied in capital proceedings.
Petitioner protests that to read § 2264(a) as supplanting §§ 2254(b) and (c) would produce “outlandish” results, Brief for Petitioner 26, a conclusion that the Court finds plausible, ante, at 333-334, and n. 7 (although it ultimately assumes otherwise). The result would have to be “outlandish,” indeed, before a court should refuse to apply the language chosen by Congress, but no such result would obtain here. Petitioner and the Court both fail to appreciate the different litigating incentives facing capital and noncapital defendants. Noncapital defendants, serving criminal sentences in prison, file habeas petitions seeking to be released, presumably as soon as possible. They have no incentive to delay. In such circumstances, §§ 2254(b) and (c) quite reasonably require that their habeas claims be filed first in state courts, so that the state judicial apparatus may have the first opportunity to address those claims. In contrast, capital defendants, facing impending execution, seek to avoid being executed. Their incentive, therefore, is to utilize every means possible to delay the carrying out of their sentence. It is, therefore, not at all “outlandish” for Congress to have concluded that in such circumstances §§ 2254(b) and (c) exhaustion would needlessly prolong capital proceedings and that § 2264(a)’s requirement that a claim have been raised and decided on the merits in state court was a sufficient protection of States’ interests in exhaustion.2
*341At this point the Court’s analysis stops. Based on the weak inference from Congress’ designation of chapter 154 as applying to pending cases and a strained reading of §2264, the Court concludes that Congress impliedly intended for chapter 153 not to apply to pending cases. I would go on, and apply our ordinary retroactivity principles, as Congress no doubt assumed that we would.
First, we have generally applied new procedural rules to pending cases. Landgraf, 511 U. S., at 275; see also Beazell v. Ohio, 269 U. S. 167, 170-171 (1925); Ex parte Collett, 337 U. S. 55, 71 (1949); Dobbert v. Florida, 432 U. S. 282, 293-294 (1977); Collins v. Youngblood, 497 U. S. 37, 45 (1990). This is because “rules of procedure regulate secondary rather than primary conduct.” Landgraf, supra, at 275. Here, the primary conduct occurred when Lindh murdered two people in the sheriff’s office of the City-County Building in Madison, Wisconsin. Obviously, the AEDPA in no way purports to regulate that past conduct. Lindh’s state-court proceedings constituted secondary conduct. Under our retroactivity *342precedents, were his state proceedings in federal court, we would have then applied existing procedural law, even though Lindh’s primary conduct occurred some time earlier. The federal habeas proceeding at issue here is, in a sense, tertiary conduct. It is not the actual criminal conduct prohibited by law, nor is it the proceeding to determine whether the defendant in fact committed such conduct. Rather, it is a collateral proceeding that, in effect, attacks the judgment of the prior state proceeding. Section 2254(d), the precise section at issue here, simply alters the standard under which that prior judgment is evaluated, and is in that sense entirely procedural. Cf. Horning v. District of Columbia, 254 U. S. 135, 139 (1920) (applying newly enacted harmless-error statute, which changed the standard under which prior judgments were evaluated, to pending case).
Second, we have usually applied changes in law to prospective forms of relief. Landgraf supra, at 273; see also Duplex Printing Press Co. v. Deering, 254 U. S. 443, 464 (1921); American Steel Foundries v. Tri-City Central Trades Council, 257 U. S. 184, 201 (1921); Hall v. Beals, 396 U. S. 45, 48 (1969) (per curiam); Kaiser Aluminum & Chemical Corp. v. Bonjorno, 494 U. S. 827, 852 (1990) (Scalia, J., concurring). Unlike damages actions, which are “quintessential^ backward looking,” Landgraf, supra, at 282, the writ of habeas corpus is prospective in nature. Habeas does not compensate for past wrongful incarceration, nor does it punish the State for imposing it. See Lane v. Williams, 455 U. S. 624, 631 (1982). Instead, habeas is a challenge to unlawful custody, and when the writ issues it prevents further illegal custody. See Preiser v. Rodriguez, 411 U. S. 475, 489, 494 (1973).
Finally, we have regularly applied statutes ousting jurisdiction to pending litigation.3 Landgraf, supra, at 274; see *343also Bruner v. United States, 343 U. S. 112, 116-117, and n. 8 (1952) (“Congress has not altered the nature or validity of petitioner’s rights or the Government’s liability but has simply reduced the number of tribunals authorized to hear and determine such rights and liabilities”); Hallowell v. Commons, 239 U. S. 506, 508 (1916); Sherman v. Grinnell, 123 U. S. 679, 680 (1887); Assessors v. Osbornes, 9 Wall. 567, 575 (1870); Ex parte McCardle, 7 Wall. 506, 514 (1869); Insurance Co. v. Ritchie, 5 Wall. 541, 544-545 (1867). This is because such statutes “ ‘speak to the power of the court rather than to the rights or obligations of the parties.’” Landgraf, supra, at 274 (quoting Republic Nat. Bank of Miami v. United States, 506 U. S. 80, 100 (1992) (Thomas, J., concurring)); see also 511 U. S., at 293 (Scalia, J., concurring in judgment) (“Our jurisdiction cases are explained, I think, by the fact that the purpose of provisions conferring or eliminating jurisdiction is to permit or forbid the exercise of judicial-power — so that the relevant event for retroactivity purposes is the moment at which that power is sought to be exercised”). This is the principle most relevant to the case at hand.
There is a good argument that § 2254(d) is itself jurisdictional. See Brown v. Allen, 344 U. S. 443, 460 (1953) (“Jurisdiction over applications for federal habeas corpus is controlled by statute”); Sumner v. Mata, 449 U. S. 539, 547, n. 2 (1981) (“The present codification of the federal habeas statute is the successor to ‘the first congressional grant of jurisdiction to the federal courts,’ and the 1966 amendments embodied in § 2254(d) [now codified, as amended by the AEDPA, at § 2254(e)] were intended by Congress as limitations on the exercise of that jurisdiction” (quoting Preiser v. Rodriguez, supra, at 485)); cf. Arkansas v. Farm Credit Servs. of Central Ark., 520 U. S. 821, 826 (1997) (explaining that the Tax Injunction Act — which has operative language similar to *344§ 2254(d) (“The district courts shall not enjoin ...”) — is “first and foremost a vehicle to limit drastically federal district court jurisdiction to interfere with so important a local concern as the collection of taxes” (internal quotation marks omitted)). But even if it is not jurisdictional, it sháres the most salient characteristic of jurisdictional statutes: Its commands are addressed to courts rather than to individuals. Section 2254(d) does not address criminal defendants, or even state prosecutors; it prescribes or proscribes no private conduct. Instead, it is addressed directly to federal courts, providing, “[a]n application for a writ of habeas corpus on behalf of a person in custody pursuant to the judgment of a State court shall not be granted . . . unless . . . .” (Emphasis added.)
Whether the approach is framed in terms of “retroactive effect,” as the Landgraf majority put it, 511 U. S., at 280, or in terms of “the relevant activity that the rule regulates,” as Justice Scalia’s concurrence put it, see id., at 291 (opinion concurring in judgment), our longstanding practice of applying procedural, prospective, and jurisdiction-ousting statutes to pending cases must play an important part in the decision. These principles all favor application of § 2254(d) to pending cases.
It is a procedural statute, regulating prospective relief, and addressed directly to federal courts and removing their power to give such relief in specified circumstances. Our cases therefore strongly suggest that, absent congressional direction otherwise, we should apply § 2254(d) to pending cases. This is not because of any peculiar characteristic intrinsic to the writ of habeas corpus, but rather because modifications to federal courts’ authority to issue the writ are necessarily of that stripe — procedural, prospective, and addressed to courts. It is therefore not surprising that the parties have not pointed us to a single case where we have found a modification in the scope of habeas corpus relief inapplicable to pending cases. To the contrary, respondent and *345amici have pointed instead to the uniform body of our eases applying such changes to all pending cases. This has been true both of statutory changes in the scope of the writ, see, e. g., Gusik v. Schilder, 340 U. S. 128, 131-133, and n. 4 (1950) (applying 1948 habeas amendments to pending claims); Smith v. Yeager, 393 U. S. 122, 124-125 (1968) (per curiam) (applying 1966 habeas amendments to pending claims); Carafas v. LaVallee, 391 U. S. 234, 239 (1968) (same); Felker v. Turpin, 518 U. S. 651 (1996) (applying different section of the AEDPA to pending case), and of judicial changes, see, e. g., Stone v. Powell, 428 U. S. 465, 495, n. 38 (1976) (rejecting petitioner’s contention that change in law should apply prospectively); Sumner v. Mata, supra, at 539, 549-551 (applying presumption of correctness of state-court findings of fact to pending case); Wainwright v. Sykes, 433 U. S. 72 (1977) (applying the cause and prejudice doctrine to pending case); Brecht v. Abrahamson, 507 U. S. 619, 638-639 (1993) (applying actual prejudice standard to pending case).
Because the Court’s inquiry is incomplete, I believe it has reached the wrong result in this case. I would affirm the judgment of the Court of Appeals.
See, e. g., Pet. for Habeas Corpus in In re Mata, O. T. 1995, No. 96-5679, p. 7 (describing how it took nine years and three months for a Federal District Court to deny, and the Ninth Circuit to affirm, petitioner’s first federal capital habeas petition).
This conclusion would also be consistent with the conclusions of the Powell Committee, which was convened to address the problems in capital habeas eases and upon whose recommendations chapter 154 was substan*341tially based. See Judicial Conference of the United States, Ad Hoc Committee on Federal Habeas Corpus in Capital Cases, Committee Report and Proposal (Aug. 23, 1989). The Committee’s Comment to Proposed § 2259 (which tracks the AEDPA’s § 2264) explained as follows: “As far as new or ‘unexhausted’ claims are concerned, [this section] represents a change in the exhaustion doctrine as articulated in Rose v. Lundy, 455 U. S. 509 (1982). [This section] bars such claims from consideration unless one of the . . . exceptions is applicable. The prisoner cannot return to state court to exhaust even if he would like to do so. On the other hand, if [an exception] is applicable, the district court is directed to conduct an evidentiary hearing and to rule on the new claim without first exhausting state remedies as Rose v. Lundy now requires. Because of the existence of state procedural default rules, exhaustion is futile in the great majority of cases. It serves the state interest of comity in theory, but in practice it results in delay and undermines the state interest in the finality of its criminal convictions. The Committee believes that the States would prefer to see post-conviction litigation go forward in capital cases, even if that entails a minor subordination of their interest in comity as it is expressed in the exhaustion doctrine.” Id., at 22-23 (emphasis added).
Although in Hughes Aircraft Co. v. United States ex rel. Schumer, 520 U. S. 939 (1997), we recently rejected a presumption favoring retroactivity for jurisdietion-creaimp statutes, see id., at 950-951, nothing in Hughes *343disparaged our longstanding practice of applying jurisdiction-oMSimgr statutes to pending cases.