delivered the opinion of the Court.

Both the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), enacted on April 24,1996, 110 Stat. 1214, and the Illegal Immigration Reform and Immigrant Responsibility Act of 1996 (IIRIRA), enacted on September 30, 1996, 110 Stat. 3009-546, contain comprehensive amendments to the Immigration and Nationality Act (INA), 66 Stat. 163, as amended, 8 U. S. C. § 1101 et seq. This case raises two important questions about the impact of those amendments. The first question is a procedural one, concerning the effect of those amendments on the availability of habeas corpus jurisdiction under 28 U. S. C. §2241. The second question is a • substantive one, concerning the impact of the amendments on conduct that occurred before *293their enactment and on the availability of discretionary relief from deportation.

Respondent, Enrico St. Cyr, is a citizen of Haiti who was admitted to the United States as a lawful permanent resident in 1986. Ten years later, on March 8, 1996, he pleaded guilty in a state court to a charge of selling a controlled substance in violation of Connecticut law. That conviction made him deportable. Under pre-AEDPA law applicable at the time of his conviction, St. Cyr would have been eligible for a waiver of deportation at the discretion of the Attorney General. However, removal proceedings against him were not commenced until April 10, 1997, after both AEDPA and IIRIRA became effective, and, as the Attorney General interprets those statutes, he no longer has discretion to grant such a waiver.

In his habeas corpus petition, respondent has alleged that the restrictions on discretionary relief from deportation contained in the 1996 statutes do not apply to removal proceedings brought against an alien who pleaded guilty to a de-portable crime before their enactment. The District Court accepted jurisdiction of his application and agreed with his submission. In accord with the decisions of four other Circuits, the Court of Appeals for the Second Circuit affirmed.1 229 F. 3d 406 (2000). The importance of both questions warranted our grant of certiorari. 531 U. S. 1107 (2001).

I

The character of the pre-AEDPA and pre-IIRIRA law that gave the Attorney General discretion to waive deportation in certain cases is relevant to our appraisal of both the substantive and the procedural questions raised by *294the petition of the Immigration and Naturalization Service (INS). We shall therefore preface our discussion of those questions with an overview of the sources, history, and scope of that law.

Subject to certain exceptions, § 3 of the Immigration Act of 1917 excluded from admission to the United States several classes of aliens, including, for example, those who had committed crimes “involving moral turpitude.” 39 Stat. 875. The seventh exception provided “[t]hat aliens returning after a temporary absence to an unrelinquished United States domicile of seven consecutive years may be admitted in the discretion of the Secretary of Labor, and under such conditions as he may prescribe.” Id., at 878.2 Although that provision applied literally only to exclusion proceedings, and although the deportation provisions of the statute did not contain a similar provision, the INS relied on §3 to grant relief in deportation proceedings involving aliens who had departed and returned to this country after the ground for deportation arose. See, e. g., Matter of L, 1 I. & N. Dec. 1, 2 (1940).3

Section 212 of the Immigration and Nationality Act of 1952, which replaced and roughly paralleled § 3 of the 1917 Act, excluded from the United States several classes of aliens, including those convicted of offenses involving moral turpitude or the illicit traffic in narcotics. See 66 Stat. 182-187. As with the prior law, this section was subject to a proviso granting the Attorney General broad discretion to *295admit excludable aliens. See id., at 187. That proviso, codified at 8 U. S. C. § 1182(c), stated:

“Aliens lawfully admitted for permanent residence who temporarily proceeded abroad voluntarily and not under an order of deportation, and who are returning to a lawful unrelinquished domicile of seven consecutive years, may be admitted in the discretion of the Attorney General....”

Like § 3 of the 1917 Act, § 212(c) was literally applicable only to exclusion proceedings, but it too has been interpreted by the Board of Immigration Appeals (BIA) to authorize any permanent resident alien with “a lawful unrelinquished domicile of seven consecutive years” to apply for a discretionary waiver from deportation. See Matter of Silva, 16 I. & N. Dec. 26, 30 (1976) (adopting position of Francis v. INS, 532 F. 2d 268 (CA2 1976)). If relief is granted, the deportation proceeding is terminated and the alien remains a permanent resident.

The extension of § 212(c) relief to the deportation context has had great practical importance, because deportable offenses have historically been defined broadly. For example, under the INA, aliens are deportable upon conviction for two crimes of “moral turpitude” (or for one such crime if it occurred within five years of entry into the country and resulted in a jail term of at least one year). See 8 U. S. C. §§ 1227(a)(2)(A)(i)-(ii) (1994 ed., Supp. V). In 1988, Congress further specified that an alien is deportable upon conviction for any “aggravated felony,” Anti-Drug Abuse Act of 1988, 102 Stat. 4469-4470, § 1227(a)(2)(A)(iii), which was defined to include numerous offenses without regard to how long ago they were committed.4 Thus, the class of aliens *296whose continued residence in this country has depended on their eligibility for § 212(c) relief is extremely large, and not surprisingly, a substantial percentage of their applications for § 212(c) relief have been granted.5 Consequently, in the period between 1989 and 1995 alone, § 212(c) relief was granted to over 10,000 aliens.6

*297Three statutes enacted in recent years have reduced the size of the class of aliens eligible for such discretionary relief. In 1990, Congress amended § 212(c) to preclude from discretionary relief anyone convicted of an aggravated felony who had served a term of imprisonment of at least five years. §511, 104 Stat. 5052 (amending 8 U. S. C. § 1182(c)). In 1996, in § 440(d) of AEDPA, Congress identified a broad set of offenses for which convictions would preclude such relief. See 110 Stat. 1277 (amending 8 U. S. C. § 1182(c)).7 And finally, that same year, Congress passed IIRIRA. That statute, inter alia, repealed § 212(c), see § 304(b), 110 Stat. 3009-597, and replaced it with a new section that gives the Attorney General the authority to cancel removal for a narrow class of inadmissible or deportable aliens, see id., at 3009-594 (creating 8 U. S. C. § 1229b (1994 ed., Supp. V)). So narrowed, that class does not include anyone previously “convicted of any aggravated felony.” §1229b(a)(3) (1994 ed., Supp. V).

In the Attorney General’s opinion, these amendments have entirely withdrawn his § 212(c) authority to waive deportation for aliens previously convicted of aggravated felonies. Moreover, as a result of other amendments adopted in AEDPA and IIRIRA, the Attorney General also maintains that there is no judicial forum available to decide whether these statutes did, in fact, deprive him of the power to grant such relief. As we shall explain below, we disagree on both points. In our view, a federal court does have jurisdiction to decide the merits of the legal question, and *298the District Court and the Court of Appeals decided that question correctly in this case.

1 — 4 K-<

The first question we must consider is whether the District Court retains jurisdiction under the general habeas corpus statute, 28 U. S. C. § 2241, to entertain St. Cyr’s challenge. His application for a writ raises a pure question of law. He does not dispute any of the facts that establish his deport-ability or the conclusion that he is deportable. Nor does he contend that he would have any right to have an unfavorable exercise of the Attorney General’s discretion reviewed in a judicial forum. Rather, he contests the Attorney General’s conclusion that, as a matter of statutory interpretation, he is not eligible for discretionary relief.

The District Court held, and the Court of Appeals agreed, that it had jurisdiction to answer that question in a habeas corpus proceeding.8 The INS argues, however, that four sections of the 1996 statutes — specifically, § 401(e) of AEDPA and three sections of IIRIRA (8 U. S. C. §§ 1252(a)(1), 1252(a)(2)(C), and 1252(b)(9) (1994 ed., Supp. V)) — stripped the, courts of jurisdiction to decide the question of law presented by respondent’s habeas corpus application.

For the INS to prevail it must overcome both the strong presumption in favor of judicial review of administrative action9 and the longstanding rule requiring a clear statement of congressional intent to repeal habeas jurisdiction. See Ex parte Yerger, 8 Wall. 85, 102 (1869) (“We are not at liberty to except from [habeas corpus jurisdiction] any cases not plainly excepted by law”); Felker v. Turpin, 518 U. S. 651, 660-661 (1996) (noting that “[n]o provision of Title I *299mentions our authority to entertain original habeas petitions,” and the statute “makes no mention of our authority to hear habeas petitions filed as original matters in this Court”).10 Implications from statutory text or legislative history are not sufficient to repeal habeas jurisdiction; instead, Congress must articulate specific and unambiguous statutory directives to effect a repeal. Ex parte Yerger, 8 Wall., at 105 (“Repeals by implication are not favored. They are seldom admitted except on the ground of repugnancy; and never, we think, when the former act can stand together with the new act”).11

In this case, the plain statement rule draws additional reinforcement from other canons of statutory construction. First, as a general matter, when a particular interpretation of a statute invokes the outer limits of Congress’ power, we expect a clear indication that Congress intended that result. See Edward J DeBartolo Corp. v. Florida Gulf Coast Building & Constr. Trades Council, 485 U. S. 568, 575 (1988). Second, if an otherwise acceptable construction of a statute *300would raise serious constitutional problems, and where an alternative interpretation of the statute is “fairly possible,” see Crowell v. Benson, 285 U. S. 22, 62 (1932), we are obligated to construe the statute to avoid such problems. See Ashwander v. TVA, 297 U. S. 288, 341, 345-348 (1936) (Brandeis, J., concurring); United States ex rel. Attorney General v. Delaware & Hudson Co., 213 U. S. 366, 408 (1909).12

A construction of the amendments at issue that would entirely preclude review of a pure question of law by any court would give rise to substantial constitutional questions. Article I, § 9, cl. 2, of the Constitution provides: “The Privilege of the Writ of Habeas Corpus shall not be suspended, unless when in Cases of Rebellion or Invasion the public Safety may require it.” Because of that Clause, some “judicial intervention in deportation cases” is unquestionably “required by the Constitution.” Heikkila v. Barber, 345 U. S. 229, 235 (1953).

Unlike the provisions of AEDPA that we construed in Felker v. Turpin, 518 U. S. 651 (1996), this case involves an alien subject to a federal removal order rather than a person confined pursuant to a state-court conviction. Accordingly, regardless of whether the protection of the Suspension *301Clause encompasses all cases covered by the 1867 Amendment extending the protection of the writ to state prisoners, cf. id., at 663-664, or by subsequent legal developments, see LaGuerre v. Reno, 164 F. 3d 1035 (CA7 1998), at the absolute minimum, the Suspension Clause protects the writ “as it existed in 1789.”13 Felker, 518 U. S., at 663-664.

At its historical core, the writ of habeas corpus has served as a means of reviewing the legality of Executive detention, and it is in that context that its protections have been strongest.14 See, e. g., Swain v. Pressley, 430 U. S. 372, 380, n. 13 (1977); id., at 385-386 (Burger, C. J., concurring) (noting that “the traditional Great Writ was largely a remedy against executive detention”); Brown v. Allen, 344 U. S. 443, 533 (1953) (Jackson, J., concurring in result) (“The historic purpose of the writ has been to relieve detention by executive authorities without judicial trial”). In England prior to 1789, in the Colonies,15 and in this Nation during the formative years of our Government, the writ of habeas corpus was available to nonenemy aliens as well as to citiz*302ens.16 It enabled them to challenge Executive and private detention in civil cases as well as criminal.17 Moreover, the issuance of the writ was not limited to challenges to the jurisdiction of the custodian, but encompassed detentions based on errors of law, including the erroneous application or interpretation of statutes.18 It was used to command the discharge of seamen who had a statutory exemption from impressment into the British Navy,19 to emancipate slaves,20 and to obtain the freedom of apprentices21 and asylum inmates.22 Most important, for our purposes, those early cases contain no suggestion that habeas relief in cases in*303volving Executive detention was only available for constitutional error.23

Notwithstanding the historical use of habeas corpus to remedy unlawful Executive action, the INS argues that this case falls outside the traditional scope of the writ at common law. It acknowledges that the writ protected an individual who was held without legal authority,' but argues that the writ would not issue where “an official had statutory authorization to detain the individual . . . but ... the official was not properly exercising his discretionary power to determine whether the individual should be released.” Brief for Respondent in Colcano-Martinez v. INS, O. T. 2000, No. 00-1011, p. 33. In this case, the INS points out, there is no dispute that the INS had authority in law to hold St. Cyr, as he is eligible for removal. St. Cyr counters that there is historical evidence of the writ issuing to redress the *304improper exercise of official discretion. See n. 23, supra; Hafetz, The Untold Story of Noncriminal Habeas Corpus and the 1996 Immigration Acts, 107 Yale L. J. 2609 (1998).

St. Cyr’s constitutional position also finds some support in our prior immigration cases. In Heikkila v. Barber, the Court observed that the then-existing statutory immigration scheme “had the effect of precluding judicial intervention in deportation cases except insofar as it was required by the Constitution,” 345 U. S., at 234-236 (emphasis added) — and that scheme, as discussed below, did allow for review on habeas of questions of law concerning an alien’s eligibility for discretionary relief. Therefore, while the INS’ historical arguments are not insubstantial, the ambiguities in the scope of the exercise of the writ at common law identified by St. Cyr, and the suggestions in this Court’s prior decisions as to the extent to which habeas review could be limited consistent with the Constitution, convince us that the Suspension Clause questions that would be presented by the INS’ reading of the immigration statutes before us are difficult and significant.24

In sum, even assuming that the Suspension Clause protects only the writ as it existed in 1789, there is substantial *305evidence to support the proposition that pure questions of law like the one raised by the respondent in this case could have been answered in 1789 by a common-law judge with power to issue the writ of habeas corpus. It necessarily follows that a serious Suspension Clause issue would be presented if we were to accept the INS’ submission that the 1996 statutes have withdrawn that power from federal judges and provided no adequate substitute for its exercise. See Hart, The Power of Congress to Limit the Jurisdiction of Federal Courts: An Exercise in Dialectic, 66 Harv. L. Rev. 1362, 1395-1397 (1953). The necessity of resolving such a serious and difficult constitutional issue — and the desirability of avoiding that necessity — simply reinforce the reasons for requiring a clear and unambiguous statement of congressional intent.

Moreover, to conclude that the writ is no longer available in this context would represent a departure from historical practice in immigration law. The writ of habeas corpus has always been available to review the legality of Executive detention. See Felker, 518 U. S., at 663; Swain v. Pressley, 430 U. S., at 380, n. 13; id., at 385-386 (Burger, C. J., concurring); Brown v. Allen, 344 U. S., at 533 (Jackson, J., concurring in result). Federal courts have been authorized to issue writs of habeas corpus since the enactment of the Judiciary Act of 1789, and §2241 of the Judicial Code provides that federal judges may grant the writ of habeas corpus on the application of a prisoner held “in custody in violation of the Constitution or laws or treaties of the United States.”25 28 U. S. C. § 2241. Before and after the enactment in 1875 of the first statute regulating immigration, 18 Stat. 477, that jurisdiction was regularly invoked on behalf of noncitizens, particularly in the immigration context. See, e. g., In re *306Kaine, 14 How. 103 (1853); United States v. Jung Ah Lung, 124 U. S. 621, 626-632 (1888).

Until the enactment of the 1952 Immigration and Nationality Act, the sole means by which an alien could test the legality of his or her deportation order was by bringing a habeas corpus action in district court.26 See, e. g., United States v. Jung Ah Lung, 124 U. S. 621 (1888); Heikkila, 345 U. S., at 235; Chin Yow v. United States, 208 U. S. 8 (1908); Ng Fung Ho v. White, 259 U. S. 276, 284 (1922). In such cases, other than the question whether there was some evidence to support the order,27 the courts generally did not review factual determinations made by the Executive. See Ekiu v. United States, 142 U. S. 651, 659 (1892). However, they did review the Executive’s legal determinations. See Gegiow v. Uhl, 239 U. S. 3, 9 (1915) (“The statute by enumerating the conditions upon which the allowance to land may be denied, prohibits the denial in other cases. And when the record shows that a commissioner of immigration is exceeding his power, the alien may demand his release upon habeas corpus”); see also Neuman, Jurisdiction and the Rule of Law after the 1996 Immigration Act, 113 Harv. L. Rev. 1963, 1965-1969 (2000).28 In case after case, courts answered questions of law in ha-*307beas corpus proceedings brought by aliens challenging Executive interpretations of the immigration laws.29

Habeas courts also regularly answered questions of law that arose in the context of discretionary relief. See, e. g., United States ex rel. Accardi v. Shaughnessy, 347 U. S. 260 (1964); United States ex rel. Hintopoulos v. Shaughnessy, 353 U. S. 72, 77 (1957).30 Traditionally, courts recognized a distinction between eligibility for discretionary relief, on the one hand, and the favorable exercise of discretion, on the other.hand. See Neuman, 113 Harv. L. Rev., at 1991 (noting the “strong tradition in habeas corpus law . . . that subjects the legally erroneous failure to exereise discretion, unlike a substantively unwise exercise of discretion, to inquiry on the writ”). Eligibility that was “governed by spe*308cific statutory standards” provided “a right to a ruling on an applicant’s eligibility,” even though the actual granting of relief was “not a matter of right under any circumstances, but rather is in all cases a matter of grace.” Jay v. Boyd, 351 U. S. 345, 353-354 (1956). Thus, even though the actual suspension of deportation authorized by § 19(c) of the Immigration Act of 1917 was a matter of grace, in United States ex rel. Accardi v. Shaughnessy, 347 U. S. 260 (1954), we held that a deportable alien had a right to challenge the Executive’s failure to exercise the discretion authorized by the law. The exercise of the District Court’s habeas corpus jurisdiction to answer a pure question of law in this case is entirely consistent with the exercise of such jurisdiction in Accardi. See also United States ex rel. Hintopoulos v. Shaughnessy, 353 U. S., at 77.

Thus, under the pre-1996 statutory scheme — and consistent with its common-law antecedents — it is clear that St. Cyr could have brought his challenge to the BIA’s legal determination in a habeas corpus petition under 28 U. S. C. § 2241. The INS argues, however, that AEDPA and IIRIRA contain four provisions that express a clear and unambiguous statement of Congress’ intent to bar petitions brought under §2241, despite the fact that none of them mention that section. The first of those provisions is AEDPA’s § 401(e).

While the title of § 401(e) — “Elimination of Custody Review by Habeas Corpus” — would seem to support the INS’ submission, the actual text of that provision does not.31 As we have previously noted, a title alone is not controlling. Pennsylvania Dept. of Corrections v. Yeskey, 524 U. S. 206, *309212 (1998) (“ ‘[T]he title of a statute . . . cannot limit the plain meaning of the text. For interpretive purposes, [it is] of use only when [it] shed[s] light on some ambiguous word or phrase’ ” (quoting Trainmen v. Baltimore & Ohio R. Co., 331 U. S. 519, 528-529 (1947))). The actual text of § 401(e), unlike its title, merely repeals a subsection of the 1961 statute amending the judicial review provisions of the 1952 Immigration and Nationality Act. See n. 31, supra. Neither the title nor the text makes any mention of 28 U. S. C. §2241.

Under the 1952 Act, district courts had broad authority to grant declaratory and injunctive relief in immigration cases, including orders adjudicating deportability and those denying suspensions of deportability. See Foti v. INS, 375 U. S. 217, 225-226 (1963). The 1961 Act withdrew that jurisdiction from the district courts and provided that the procedures set forth in the Hobbs Act would be the "sole and exclusive procedure” for judicial review of final orders of deportation, subject to a series of exceptions. See 75 Stat. 651. The last of those exceptions stated that "any alien held in custody pursuant to an order of deportation may obtain review thereof by habeas corpus proceedings.” See id., at 652, codified at 8 U. S. C. § 1105a(10) (repealed Sept. 30, 1996).

The INS argues that the inclusion of that exception in the 1961 Act indicates that Congress must have believed that it would otherwise have withdrawn the pre-existing habeas corpus jurisdiction in deportation cases, and that, as a result, the repeal of that exception in AEDPA in 1996 implicitly achieved that result. It seems to us, however, that the 1961 exception is best explained as merely confirming the limited scope of the new review procedures. In fact, the 1961 House Report provides that this section "in no way disturbs the Habeas Corpus Act.”32 H. R. Rep. No. 1086, 87th Cong., 1st *310Sess., 29 (1961). Moreover, a number of the courts that considered the interplay between the general habeas provision and INA § 106(a)(10) after the 1961 Act and before the enactment of AEDPA did not read the 1961 Act’s specific habeas provision as supplanting jurisdiction under §2241. Orozco v. INS, 911 F. 2d 539, 541 (CA11 1990); United States ex rel. Marcello v. INS, 634 F. 2d 964, 967 (CA5 1981); Sotelo Mondragon v. Ilchert, 653 F. 2d 1254, 1255 (CA9 1980).

' In any case, whether § 106(a)(10) served as an independent grant of habeas jurisdiction or simply as an acknowledgment of continued jurisdiction pursuant to §2241, its repeal cannot be sufficient to eliminate what it did not originally grant— namely, habeas jurisdiction pursuant to 28 U. S. C. §2241.33 See Ex parte Yerger, 8 Wall., at 105-106 (concluding that the repeal of “an additional grant of jurisdiction” does not “operate as a repeal of jurisdiction theretofore allowed”); Ex parte McCardle, 7 Wall. 506, 515 (1869) (concluding that the repeal of portions of the 1867 statute conferring appellate jurisdiction on the Supreme Court in habeas proceedings did “not affect the jurisdiction which was previously exercised”).

The INS also relies on three provisions of IIRIRA, now codified at 8 U. S. C. §§ 1252(a)(1), 1252(a)(2)(C), and *3111252(b)(9) (1994 ed., Supp. V). As amended by §306 of IIRIRA, 8 U. S. C. § 1252(a)(1) (1994 ed., Supp. V) now provides that, with certain exceptions, including those set out in subsection (b) of the same statutory provision, “[j]udicial review of a final order of removal... is governed only by” the Hobbs Act’s procedures for review of agency orders in the courts of appeals. Similarly, § 1252(b)(9), which addresses the “ [consolidation of questions for judicial review,” provides that “[judicial review of all questions of law and fact, including interpretation and application of constitutional and statutory provisions, arising from any action taken or proceeding brought to remove an alien from the United States under this subchapter shall be available only in judicial review of a final order under this section.”34 Finally, § 1252(a)(2)(C), which concerns “[mjatters not subject to judicial review,” states: “Notwithstanding any other provision of law, no court shall have jurisdiction to review any final order of removal against an alien who is removable by reason of having committed” certain enumerated criminal offenses.

The term “judicial review” or “jurisdiction to review” is the focus of each of these three provisions. In the immigration context, “judicial review” and “habeas corpus” have historically distinct meanings. See Heikkila v. Barber, 345 U. S. 229 (1953). In Heikkila, the Court concluded that the finality provisions at issue “precluded] judicial review” to the maximum extent possible under the Constitution, and thus concluded that the APA was inapplicable. Id., at 235. Nevertheless, the Court reaffirmed the right to habeas *312corpus. Ibid. Noting that the limited role played by the courts in habeas corpus proceedings was far narrower than the judicial review authorized by the APA, the Court concluded that “it is the scope of inquiry on habeas corpus that differentiates” habeas review from “judicial review.” Id., at 236; see also, e. g., Terlinden v. Ames, 184 U. S. 270, 278 (1902) (noting that under the extradition statute then in effect there was “no right of review to be exercised by any court or judicial officer,” but that limited review on habeas was nevertheless available); Ekiu, 142 U. S., at 663 (observing that while a decision to exclude an alien was subject to inquiry on habeas, it could not be “impeached or reviewed”). Both §§ 1252(a)(1) and (a)(2)(C) speak of “judicial review”— that is, full, nonhabeas review. Neither explicitly mentions habeas,35 or 28 U. S. C. § 2241.36 Accordingly, neither pro*313vision speaks with sufficient clarity to bar jurisdiction pursuant to the general habeas statute.

The INS also makes a separate argument based on 8 U. S. C. § 1252(b)(9) (1994 ed., Supp. V). We have previously described § 1252(b)(9) as a “zipper clause.” A ADC, 525 U. S. 471, 483 (1999). Its purpose is to consolidate “judicial review” of immigration proceedings into one action in the court of appeals, but it applies only “[wjith respect to review of an order of removal under subsection (a)(1).” 8 U. S. C. § 1252(b) (1994 ed., Supp. V).37 Accordingly, this provision, by its own terms, does not bar habeas jurisdiction over removal orders not subject to judicial review under § 1252(a)(1) — including orders against aliens who are removable by reason of having committed one or more criminal offenses. Subsection (b)(9) simply provides for the consolidation of issues to be brought in petitions for “[jjudicial review,” which, as we note above, is a term historically dis*314tinct from habeas. See Mahadeo v. Reno, 226 F. 3d 3, 12 (CA1 2000); Flores-Miramontes v. INS, 212 F. 3d 1133, 1140 (CA9 2000). It follows that § 1252(b)(9) does not clearly apply to actions brought pursuant to the general habeas statute, and thus cannot repeal that statute either in part or in whole.

If it were clear that the question of law could be answered in another judicial forum, it might be permissible to accept the INS’ reading of § 1252. But the absence of such a forum, coupled with the lack of a clear, unambiguous, and express statement of congressional intent to preclude judicial consideration on habeas of such an important question of law, strongly counsels against adopting a construction that would raise serious constitutional questions.38 Cf. Felker, 518 U. S., at 660-661. Accordingly, we conclude that habeas jurisdiction under §2241 was not repealed by AEDPA and IIRIRA.

Ill

The absence of a clearly expressed statement of congressional intent also pervades our review of the merits of St. Cyr’s claim. Two important legal consequences ensued from respondent’s entry of a guilty plea in March 1996: (1) He became subject to deportation, and (2) he became eligible for a discretionary waiver of that deportation under the pre*315vailing interpretation of § 212(c). When IIRIRA went into effect in April 1997, the first consequence was unchanged except for the fact that the term “removal” was substituted for “deportation.” The issue that remains to be resolved is whether IIRIRA § 304(b) changed the second consequence by eliminating respondent’s eligibility for a waiver.

The INS submits that the statute resolves the issue because it unambiguously communicates Congress’ intent to apply the provisions of IIRIRA’s Title III-A to all removals initiated after the effective date of the statute, and, in any event, its provisions only operate prospectively and not retrospectively. ,The Court of Appeals, relying primarily on the analysis in our opinion in Landgraf v. USI Film Products, 511 U. S. 244 (1994), held, contrary to the INS’ arguments, that Congress’ intentions concerning the application of the “Cancellation of Removal” procedure are ambiguous and that the statute imposes an impermissible retroactive effect on aliens who, in reliance on the possibility of § 212(c) relief, pleaded guilty to aggravated felonies. See 229 F. 3d, at 416, 420. We agree.

Retroactive statutes raise special concerns. See Land-graf, 511 U. S., at 266. “The Legislature’s unmatched powers allow it to sweep away settled expectations suddenly and without individualized consideration. Its responsivity to political pressures poses a risk that it may be tempted to use retroactive legislation as a means of retribution against unpopular groups or individuals.”39 Ibid. Accordingly, “congressional enactments . . . will not be construed to have retroactive effect unless their language requires this *316result.” Bowen v. Georgetown Univ. Hospital, 488 U. S. 204, 208 (1988).

“[This] presumption against retroactive legislation is deeply rooted in our jurisprudence, and embodies a legal doctrine centuries older than our Republic. Elementary considerations of fairness dictate that individuals should have an opportunity to know what the law is and to conform their conduct accordingly; settled expectations should not be lightly disrupted. For that reason, the 'principle that the legal effect of conduct should ordinarily be assessed under the law that existed when the conduct took place has timeless and universal human appeal.’ Kaiser, 494 U. S., at 855 (Scalia, J., concurring). In a free, dynamic society, creativity in both commercial and artistic endeavors is fostered by a rule of law that gives people confidence about the legal consequences of their actions.” Landgraf, 511 U. S., at 265-266 (footnote omitted).

Despite the dangers inherent in retroactive legislation, it is beyond dispute that, within constitutional limits, Congress has the power to enact laws with retrospective effect. See id., at 268. A statute may not be applied retroactively, however, absent a clear indication from Congress that it intended such a result. “Requiring clear intent assures that Congress itself has affirmatively considered the potential unfairness of retroactive application and determined that it is an acceptable price to pay for the countervailing benefits.” Id., at 272-273. Accordingly, the first step in determining whether a statute has an impermissible retroactive effect is to ascertain whether Congress has directed with the requisite clarity that the law be applied retrospectively. Martin v. Radix, 527 U. S. 343, 352 (1999).

The standard for finding such unambiguous direction is a demanding one.. “[C]ases where this Court has found truly 'retroactive’ effect adequately authorized by statute have *317involved statutory language that was so clear that it could sustain only one interpretation.” Lindk v. Murphy, 521 U. S. 820, 328, n. 4 (1997). The INS makes several arguments in favor of its position that IIRIRA achieves this high level of clarity.

First, the INS points to the comprehensive nature of IIRIRA’s revision of federal immigration law. “Congress’s comprehensive establishment of a new immigration framework,” the INS argues, “shows its intent that, after a transition period, the provisions of the old law should no longer be applied at all.” Brief for Petitioner 33-34. We rejected a similar argument, however, in Landgraf, a case that, like this one, involved Congress’ comprehensive revision of an important federal statute. 511 U. S., at 260-261. By itself, the comprehensiveness of a congressional enactment says nothing about Congress’ intentions with respect to the retro-activity of the enactment’s individual provisions.40

The INS also points to the effective date for Title III-A as providing a clear statement of congressional intent to apply IIRIRA’s repeal of § 212(c) retroactively. See IIRIRA § 309(a), 110 Stat. 3009-625. But the mere promulgation of an effective date for a statute does not provide sufficient assurance that Congress specifically considered the potential unfairness that retroactive application would produce. For that reason, a “statement that a statute will become effective on a certain date does not even arguably suggest that it has any application to conduct that occurred at an earlier date.” Landgraf, 511 U. S., at 257.

The INS further argues that any ambiguity in Congress’ intent is wiped away by the “saving provision” in IIRIRA § 309(c)(1), 110 Stat. 3009-625. Brief for Petitioner 34-36. That provision states that, for aliens whose exclusion or deportation proceedings began prior to the Title III-A efifec-*318tive date, “the amendments made by [Title III-A] shall not apply, and ... the proceedings (including judicial review thereof) shall continue to be conducted without regard to such amendments.”41 This rule, however, does not communicate with unmistakable clarity Congress’ intention to apply its repeal of § 212(c) retroactively. Nothing in either § 309(c)(1) or the statute’s legislative history even discusses the effect of the statute on proceedings based on pre-IIRIRA convictions that are commenced after its effective date.42 Section 309(c)(1) is best read as merely setting out the procedural rules to be applied to removal proceedings pending on the effective date of the statute. Because “[cjhanges in procedural rules may often be applied in suits arising before their enactment without raising concerns about retroactivity,” Landgraf 511 U. S., at 275, it was necessary for Congress to identify which set of procedures would apply in those circumstances. As the Conference Report expressly explained, “[§ 309(c)] provides for the transition to new procedures in the case of an alien already in exclusion or deportation proceedings on the effective date.” H. R. Conf. Rep. No. 104-828, p. 222 (1996) (emphasis added).

Another reason for declining to accept the INS’ invitation to read § 309(c)(1) as dictating the temporal reach of IIRIRÁ § 304(b) is provided by Congress’ willingiiess, in other sections of IIRIRA, to indicate unambiguously its intention *319to apply specific provisions retroactively. IIRIRA’s amendment of the definition of “aggravated felony,” for example, clearly states that it applies with respect to “convietion[s]... entered before, on, or after” the statute’s enactment date. § 321(b).43 As the Court of Appeals noted, the fact that Con*320gress made some provisions of IIRIRA expressly applicable to prior convictions, but did not do so in regard to § 304(b), is an indication “that Congress did not definitively decide the issue of §304’s retroactive application to pre-enactment convictions.” See 229 F. 3d, at 415. The “saving provision” is therefore no more significant than the specification of an effective date.

The presumption against retroactive application of ambiguous statutory provisions, buttressed by “the longstanding principle of construing any lingering ambiguities in deportation statutes in favor of the alien,” INS v. Cardoza-Fonseca, 480 U. S. 421, 449 (1987), forecloses the conclusion that, in enacting § 304(b), “Congress itself has affirmatively considered the potential unfairness of retroactive application and determined that it is an acceptable price to pay for the countervailing benefits.”44 Landgraf 511 U. S., at 272-273. We therefore proceed to the second step of Landgraf s retro-activity analysis in order to determine whether depriving removable aliens of consideration for § 212(c) relief produces an impermissible retroactive effect for aliens who, like respondent, were convicted pursuant to a plea agreement at a time when their plea would not have rendered them ineligible for § 212(c) relief.45

*321“The inquiry into whether a statute operates retroactively demands a commonsense, functional judgment about ‘whether the new provision attaches new legal consequences to events completed before its enactment.’” Martin, 527 U. S., at 357-358 (quoting Landgraf, 511 U. S., at 270). A statute has retroactive effect when it “‘takes away or impairs vested rights acquired under existing laws, or creates a new obligation, imposes a new duty, or attaches a new disability, in respect to transactions or considerations already past46 Id., at 269 (quoting Society for Propagation of the Gospel v. Wheeler, 22 F. Cas. 756, 767 (No. 13,156) (CC NH 1814) (Story, J.)). As we have repeatedly counseled, the judgment whether a particular statute acts retroactively “should be informed and guided by ‘familiar considerations of fair notice, reasonable reliance, and settled expectations.’” Martin, 527 U. S., at 358 (quoting Landgraf, 511 U. S., at 270).

IIRIRA’s elimination of any possibility of § 212(c) relief for people who entered into plea agreements with the expectation that they would be eligible for such relief clearly “‘attaches a new disability, in respect to transactions or considerations already past.’” Id., at 269. Plea agreements involve a quid pro quo between a criminal defendant and the government. See Newton v. Rumery, 480 U. S. 386, *322393, n. 3 (1987). In exchange for some perceived benefit, defendants waive several of their constitutional rights (including the right to a trial) and grant the government numerous “tangible benefits, such as promptly imposed punishment without the expenditure of prosecutorial resources. ” 47 Ibid. There can be little doubt that, as a general matter, alien defendants considering whether to enter into a plea agreement are acutely aware of the immigration consequences of their convictions.48 See Magana-Pizano v. INS, 200 F. 3d 603, 612 (CA9 1999) (“That an alien charged with a crime . . . would factor the immigration consequences of conviction in deciding whether to plead or proceed to trial is well-documented”); see also 3 Bender, Criminal Defense Techniques §§ 60A.01, 60A.02[2] (1999) (“ ‘Preserving the client’s right to remain in the United States may be more important to the client than any potential jail sentence’” (quoted in Brief for National Association of Criminal Defense Law*323yers et al. as Amici Curiae 13)). Given the frequency with which § 212(c) relief was granted in the years leading up to AEDPA and IIRIRA,49 preserving the possibility of such relief would have been one of the principal benefits sought by defendants deciding whether to accept a plea offer or instead to proceed to trial.50

The case of Charles Jideonwo, a petitioner in a parallel litigation in the Seventh Circuit, is instructive. Charged in 1994 with violating federal narcotics law, Jideonwo entered into extensive plea negotiations with the Government, the sole purpose of which was to ensure that ‘“he got less than five years to avoid what would have been a statutory bar on 212(c) relief.’” Jideonwo v. INS, 224 F. 3d 692, 699 (CA7 2000) (quoting the Immigration Judge’s findings of fact). The potential for unfairness in the retroactive application of IIRIRA § 304(b) to people like Jideonwo and St. Cyr is significant and manifest. Relying upon settled practice, the advice of counsel, and perhaps even assurances in open court that the entry of the plea would not foreclose § 212(c) relief, a great number of defendants in Jideonwo’s and St. Cyr’s position agreed to plead guilty.51 Now that prosecutors have received the benefit of these plea agreements, agreements that were likely facilitated by the aliens' belief in their continued eligibility for § 212(c) relief, it would surely be contrary to “familiar considerations of fair notice, reasonable reliance, and settled expectations,” Landgraf, 511 U. S., *324at 270, to hold that IIRIRA’s subsequent restrictions deprive them of any possibility of such relief.52

The INS argues that deportation proceedings (and the Attorney General’s discretionary power to grant relief from deportation) are “inherently prospective” and that, as a result, application of the law of deportation can never have a retroactive effect. Such categorical arguments are not particularly helpful in undertaking Landgrafs commonsense, functional retroactivity analysis. See Martin, 527 U. S., at 359. Moreover, although we have characterized deportation as “look[ing] prospectively to the respondent’s right to remain in this country in the future,” INS v. Lopez-Mendoza, 468 U. S. 1032, 1038 (1984), we have done so in order to reject the argument that deportation is punishment for past behavior and that deportation proceedings are therefore subject to the “various protections that apply in the context of a criminal trial.” Ibid. As our cases make clear, the presumption against retroactivity applies far beyond the confines of the criminal law. See Landgraf, 511 U. S., at 272. And our mere statement that deportation is not punishment for past crimes does not mean that we cannot consider an alien’s reasonable reliance on the continued availability of discretionary relief from deportation when deciding whether the elimination of such relief has a retroactive effect.53

*325Finally, the fact that § 212(c) relief is discretionary does not affect the propriety of our conclusion. There is a clear difference, for the purposes of retroactivity analysis, between facing possible deportation and facing certain deportation. Cf. Hughes Aircraft Co. v. United States ex rel. Schumer, 520 U. S. 939, 949 (1997) (an increased likelihood of facing a qui tarn, action constitutes an impermissible retroactive effect for the defendant); Lindsey v. Washington, 301 U. S. 397, 401 (1937) (“Removal of the possibility of a sentence of less than fifteen years . . . operates to [defendants’] detriment” (emphasis added)). Prior to AEDPA and IIRIRA, aliens like St. Cyr had a significant likelihood of receiving § 212(c) relief.54 Because respondent, and other aliens like him, almost certainly relied upon that likelihood in deciding whether to forgo their right to a trial, the elimination of any possibility of § 212(c) relief by IIRIRA has an obvious and severe retroactive effect.55

*326We find nothing in IIRIRA unmistakably indicating that Congress considered the question whether to apply its repeal of § 212(c) retroactively to such aliens. We therefore hold that § 212(c) relief remains available for aliens, like respondent, whose convictions were obtained through plea agreements and who, notwithstanding those convictions, would have been eligible for § 212(c) relief at the time of their plea under the law then in effect.

The judgment is affirmed.

It is so ordered.

dissenting.

I join Parts I and III of Justice Scalia’s dissenting opinion in this case. I do not join Part II because I believe that, assuming, arguendo, that the Suspension Clause guarantees some minimum extent of habeas review, the right asserted by the alien in this case falls outside the scope of that review for the reasons explained by Justice Scalia in Part II-B of his dissenting opinion. The question whether the Suspension Clause assures habeas jurisdiction in this particular case properly is resolved on this ground alone, and there is no need to say more.

with whom The Chief Justice and Justice Thomas join, and with whom Justice O’Connor joins as to Parts I and III, dissenting.

The Court today finds ambiguity in the utterly clear language of a statute that forbids the district court (and all *327other courts) to entertain the claims of aliens such as respondent St. Cyr, who have been found deportable by reason of their criminal acts. It fabricates a superclear statement, “magic words” requirement for the congressional expression of such an intent, unjustified in law and unparalleled in any other area of our jurisprudence. And as the fruit of its labors, it brings forth a version of the statute that affords criminal aliens more opportunities for delay-inducing judicial review than are afforded to noncriminal aliens, or even than were afforded to criminal aliens prior to this legislation concededly designed to expedite their removal. Because it is clear that the law deprives us of jurisdiction to entertain this suit, I respectfully dissent.

I

In categorical terms that admit of no exception, the Illegal Immigration Reform and Immigrant Responsibility Act of -1996 (IIRIRA), 110 Stat. 3009-546, unambiguously repeals the application of 28 U. S. C. § 2241 (the general habeas corpus provision), and of all other provisions for judicial review, to deportation challenges brought by certain kinds of criminal aliens. This would have been readily apparent to the reader, had the Court at the outset of its opinion set forth the relevant provisions of IIRIRA and of its statutory predecessor, the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), 110 Stat. 1214. I will begin by supplying that deficiency, and explaining IIRIRA’s jurisdictional scheme. It begins with what we have called a channeling or “‘zipper’ clause,” Reno v. American-Arab Anti-Discrimination Comm., 525 U. S. 471, 483 (1999)— namely, 8 U.S.C. § 1252(b)(9) (1994 ed., Supp.V). This provision, entitled “Consolidation of questions for judicial review,” provides as follows:

“Judicial review of all questions of law and fact, including interpretation and application of constitutional and statutory provisions, arising from any action taken *328or proceeding brought to remove an alien from the United States under this subchapter shall be available only in judicial review of a final order under this section.” (Emphases added.) ■

In other words, if. any review is available of any “questio[n] of law . . . arising from any action taken or proceeding brought to remove an alien from the United States under this súbchapter,” it is available “only in judicial review of a final order under this section [§ 1252].” What kind of review does that section provide? That is set forth in § 1252(a)(1), which states:

“Judicial review of a final order of removal (other than an order of removal without a hearing pursuant to [the expedited-removal provisions for undocumented aliens arriving at the border found in] section 1225(b)(1) of this title) is governed only by chapter 158 of title 28 [the Hobbs Act], except as provided in subsection (b) of this section [which modifies some of the Hobbs Act provisions] and except that the court may not order the taking of additional evidence under section 2847(c) of [Title 28].”

In other words, i/judicial review is available, it consists only of the modified Hobbs Act review specified in § 1252(a)(1).

In some cases (including, as it happens, the one before us), there can be no review at all, because IIRIRA categorically and unequivocally rules out judicial review of challenges to deportation brought by certain kinds of criminal aliens. Section 1252(a)(2)(C) provides:

“Notwithstanding any other provision of law, no court shall have jurisdiction to review any final order of removal’ against an alien who is removable by reason of having committed [one or more enumerated] criminal offense[s] [including drug-trafficking offenses of the sort of which respondent had been convicted].” (Emphases added.)

*329Finally, the pre-IIRIRA antecedent to the foregoing provisions — AEDPA § 401(e) — and the statutory background against which that was enacted, confirm that § 2241 habeas review, in the district court or elsewhere, has been unequivocally repealed. In 1961, Congress amended the Immigration and Nationality Act of 1952 (INA), 66 Stat. 163, by directing that the procedure for Hobbs Act review in the courts of appeals “shall .apply to, and shall be the sole and exclusive procedure for, the judicial review of all final orders of deportation” under the INA. 8 U. S. C. § 1105a(a) (repealed Sept. 30, 1996) (emphasis added). Like 8 U. S. C. § 1252(a)(2)(C) (1994 ed., Supp. V), this provision squarely prohibited §2241 district-court habeas review. At the same time that it enacted this provision, however, the 1961 Congress enacted a specific exception: “any alien held in custody pursuant to an order of deportation may obtain judicial review thereof by habeas corpus proceedings,” 8 U. S. C. § 1105a(a)(10) (1994 ed.). (This would of course have been surplusage had § 2241 habeas review not been covered by the “sole and exclusive procedure” provision.) Section 401(e) of AEDPA repealed this narrow exception, and there is no doubt what the repeal was thought to accomplish: the provision was entitled “Elimination of Custody Review by Habeas Corpus.” 110 Stat. 1268. It gave universal preclusive effect to the “sole and exclusive procedure” language of § 1105a(a). And it is this regime that IIRIRA has carried forward.

The Court’s efforts to derive ambiguity from this utmost clarity are unconvincing. First, the Court argues that §§ 1252(a)(2)(C) and 1252(b)(9) are not as clear as one might think — that, even though they are sufficient to repeal the jurisdiction of the courts of appeals, see Calcano-Martinez v. INS, post, at 351-352,1 they do not cover habeas jurisdiction in the district court, since, “[i]n the immigration context, ‘judicial review’ and ‘habeas corpus’ have historically dis*330tinct meanings,” ante, at 311, 312, n. 35. Of course § 1252(a)(2)(C) does not even use the term “judicial review” (it says “jurisdiction to review”) — but let us make believe it does. The Court’s contention that in this statute it does not include habeas corpus is decisively refuted by the language of § 1252(e)(2), enacted along with §§ 1252(a)(2)(C) and 1252(b)(9): “Judicial review of any determination made under section 1225(b)(1) of this title [governing review of expedited removal orders against undocumented aliens arriving at the border] is available in habeas corpus proceedings . . . .” (Emphases added.) It is hard to imagine how Congress could have made it any clearer that, when it used the term “judicial review” in IIRIRA, it included judicial review through habeas corpus. Research into the “historical” usage of the term “judicial review” is thus quite beside the point.

But the Court is demonstrably wrong about that as well. Before IIRIRA was enacted, from 1961 to 1996, the governing immigration statutes unquestionably treated “judicial review” as encompassing review by habeas corpus. As discussed earlier, 8 U. S. C. § 1105a (1994 ed.) made Hobbs Act review “the sole and exclusive procedure for, the judicial review of all Anal orders of deportation” (emphasis added), but created (in subsection (a)(10)) a limited exception for habeas corpus review. Section 1105a was entitled “ Judicial review of orders of deportation and exclusion” (emphasis added), and the exception for habeas corpus stated that “any alien held in custody pursuant to an order of deportation may obtain judicial review thereof by habeas corpus proceedings,” § 1105a(a)(10) (emphases added). Apart from this prior statutory usage, many of our own immigration eases belie the Court’s suggestion that the term “judicial review,” when used in the immigration context, does not include review by habeas corpus. See, e. g., United States v. Mendoza-Lopez, 481 U. S. 828, 836-837 (1987) (“[A]ny alien held in custody pursuant to an order of deportation may ob*331tain judicial review of that order in a habeas corpus proceeding” (emphases added)); Shaughnessy v. Pedreiro, 349 U. S. 48, 52 (1955) (“Our holding is that there is a right of judicial review of deportation orders other than by habeas corpus ...” (emphases added)); see also id., at 49.

The only support the Court offers in support of the asserted “longstanding distinction between ‘judicial review’ and ‘habeas,’ ” ante, at 312, n. 35, is language from a single opinion of this Court, Heikkila v. Barber, 345 U. S. 229 (1953).2 There, we “differentiate^]” “habeas corpus” from “judicial review as that term is used in the Administrative Procedure Act” Id., at 236 (emphasis added). But that simply asserts that habeas corpus review is different from ordinary APA review, which no one doubts. It does not assert that habeas corpus review is not judicial review at all. Nowhere does Heikkila make such an implausible contention.3

*332The Court next contends that the zipper clause, § 1252(b)(9), “by its own terms, does not bar” § 2241 district-court habeas review of removal orders, ante, at 313, because the opening sentence of subsection (b) states that “[w]ith respect to review of an order of removal under subsection (a)(1) of this section, the following requirements apply ... (Emphasis added.) But in the broad sense, § 1252(b)(9) does “apply” “to review of an order of removal under subsection (a)(1),” because it mandates that “review of all questions of law and fact. . . arising from any action taken or proceeding brought to remove an alien from the United States under this subchapter” must take place in connection with such review. This is “application” enough — and to insist that subsection (b)(9) be given effect only within the review of removal orders that takes place under subsection (a)(1), is to render it meaningless. Moreover, other of the numbered subparagraphs of subsection (b) make clear that the introductory sentence does not at all operate as a limitation upon what follows. Subsection (b)(7) specifies the procedure by which “a defendant in a criminal proceeding” charged with failing to depart after being ordered to do so may contest “the validity of [a removal] order” before trial; and subsection (b)(8) prescribes some of the prerogatives and responsibilities of the Attorney General and the alien after entry of a final removal order. These provisions have no effect if they must apply (even in the broad sense that subsection (b)(9) can be said to apply) “to review of an order of removal under subsection (a)(1).”

Unquestionably, unambiguously, and unmistakably, IIRIRA expressly supersedes §2241’s general provision for habeas jurisdiction. The Court asserts that Felker v. Turpin, 518 U. S. 651 (1996), and Ex parte Yerger, 8 Wall. 85 *333(1869), reflect a “longstanding rule requiring a clear statement of congressional intent to repeal habeas jurisdiction,” ante, at 298. They do no such thing. Those cases simply applied the general principle — not unique to habeas — that “[rjepeals by implication are not favored.” Felker, supra, at 660; Yerger, supra, at 105. Felker held that a statute which by its terms prohibited only further review by this Court (or by an en banc court of appeals) of a court-of-appeals panel’s “ ‘grant or denial of... authorization ... to file a second or successive [habeas] application,’ ” 518 U. S., at 657 (quoting 28 U.S.C. § 2244(b)(3)(E) (1994 ed., Supp. II)), should not be read to imply the repeal of this Court’s separate and distinct “authority [under 28 U. S. C. § 2241 and 28 U. S. C. § 2254 (1994 ed. and Supp. V)] to hear habeas petitions filed as original matters in this Court,” 518 U. S., at 661. Yerger held that an 1868 Act that by its terms “repealed] only so much of the act of 1867 as authorized appeals, or the exercise of appellate jurisdiction by this court,” should be read to “reach no [further than] the act of 1867,” and did not repeal by implication the appellate jurisdiction conferred by the Judiciary Act of 1789 and other pre-1867 enactments. 8 Wall., at 105. In the present case, unlike in Felker and Yerger, none of the statutory provisions relied upon — § 1252(a)(2)(C), § 1252(b)(9), or 8 U. S. C. § 1105a(a) (1994 ed.) — requires us to imply from one statutory provision the repeal of another. All by their terms prohibit the judicial review at issue in this case.

The Court insists, however, that since “[n]either [§ 1252(a)(1) nor § 1252(a)(2)(C)] explicitly mentions habeas, or 28 U. S. C. §2241,” “neither provision speaks with sufficient clarity to bar jurisdiction pursuant to the general ha-beas statute.” Ante, at 312-313. Even in those areas of our jurisprudence where we have adopted a “clear statement” rule (notably, the sovereign immunity cases to which the Court adverts, ante, at 299, n. 10), clear statement has never meant the kind of magic words demanded by the Court *334today — explicit reference to habeas or to § 2241 — rather than reference to “judicial review” in a statute that explicitly calls habeas corpus a form of judicial review. In Gregory v. Ashcroft, 501 U. S. 452, 467 (1991), we said:

“This [the Court’s clear-statement requirement] does not mean that the [Age Discrimination in Employment] Act must mention [state] judges explicitly, though it does not. Cf. Dellmuth v. Muth, 491 U. S. 223, 233 (1989) (Scalia, J., concurring). Rather, it must be plain to anyone reading the Act that it covers judges.”

In Gregory, as in United States v. Nordic Village, Inc., 503 U. S. 30, 34-35 (1992), and Atascadero State Hospital v. Scanlon, 473 U. S. 234, 241, 246 (1985), we held that the clear-statement requirement was not met, not because there was no explicit reference to the Eleventh Amendment, but because the statutory intent to eliminate state sovereign immunity was not clear. For the reasons discussed above, the intent to eliminate habeas jurisdiction in the present case is entirely clear, and that is all that is required.

It has happened before — too frequently, alas — that courts have distorted plain statutory text in order to produce a “more sensible” result. The unique accomplishment of today’s opinion is that the result it produces is as far removed from what is sensible as its statutory construction is from the language of the text. One would have to study our statute books for a long time to come up with a more unlikely disposition. By authorizing §2241 habeas review in the district court but foreclosing review in the court of appeals, see Calcano-Martinez, post, at 351-352, the Court’s interpretation routes all legal challenges to removal orders brought by criminal aliens to the district court, to be adjudicated under that court’s §2241 habeas authority, which specifies no time limits. After review by that court, criminal aliens will presumably have an appeal as of right to the court of appeals, and can then petition this Court for a writ of cer-*335tiorari. In contrast, noncriminal aliens seeking to challenge their removal orders — for example, those charged with having been inadmissible at the time of entry, with having failed to maintain their nonimmigrant status, with having procured a visa through a marriage that was not bona fide, or with having become, within five years after the date of entry, a public charge, see 8 U. S. C. §§ 1227(a)(1)(A), (a)(1)(C), (a)(1)(G), (a)(5) (1994 ;ed., Supp. V) — will still presumably be required to proceed directly to the court of appeals by way of petition for review, under the restrictive modified Hobbs Act review provisions set forth in § 1252(a)(1), including the 30-day filing deadline, see § 1252(b)(1). In fact, prior to the enactment of IIRIRA, criminal aliens also had to follow this procedure for immediate modified Hobbs Act review in the court of appeals. See 8 U. S. C. § 1105a(a) (1994 ed.). The Court has therefore succeeded in perverting a statu-' tory scheme designed to expedite the removal of criminal aliens into one that now affords them more opportunities for (and layers of) judicial review (and hence more opportunities for delay) than are afforded wow-criminal aliens — and more than were afforded criminal aliens prior to the enactment of IIRIRA.4 This outcome speaks for itself; no Congress ever imagined it.

To excuse the violence it does to the statutory text, the Court invokes the doctrine of constitutional doubt, which it asserts is raised by the Suspension Clause, U. S. Const., Art. I, § 9, cl. 2. This uses one distortion to justify another, transmogrifying a doctrine designed to maintain “a just re*336spect for the legislature,” Ex parte Randolph, 20 F. Cas. 242, 254 (No. 11,558) (CC Va. 1833) (Marshall, C. J., on circuit), into a means of thwarting the clearly expressed intent of the legislature. The doctrine of constitutional doubt is meant to effectuate, not to subvert, congressional intent, by giving ambiguous provisions a meaning that will avoid constitutional peril, and that will conform with Congress’s presumed intent not to enact measures of dubious validity. The condition precedent for application of the doctrine is that the statute can reasonably be construed to avoid the constitutional difficulty. See, e. g., Miller v. French, 530 U. S. 327, 341 (2000) (“ ‘We cannot press statutory construction “to the point of disingenuous evasion” even to avoid a constitutional question’ ” (quoting United States v. Locke, 471 U. S. 84, 96 (1985), in turn quoting George Moore Ice Cream Co. v. Rose, 289 U. S. 373, 379 (1933))); Salinas v. United States, 522 U. S. 52, 60 (1997) (quoting Seminole Tribe of Fla. v. Florida, 517 U. S. 44, 57, n. 9 (1996)). It is a device for interpreting what the statute says — not for ignoring what the statute says in order to avoid the trouble of determining whether what it says is unconstitutional. For the reasons I have set forth above, it is crystal clear that the statute before us here bars criminal aliens from obtaining judicial review, including §2241 district-court review, of their removal orders. It is therefore also crystal clear that the doctrine of constitutional doubt has no application.

In the remainder of this opinion I address the question the Court should have addressed: Whether these provisions of IIRIRA are unconstitutional.

II

A

The SuspensionClause of the Constitution, Art. I, § 9, cl. 2, provides as follows:

*337“The Privilege of the Writ of Habeas Corpus shall not be suspended, unless when in Cases of Rebellion or Invasion the public Safety may require it.”

A straightforward reading of this text discloses that it does not guarantee any content to (or even the existence of) the writ of habeas corpus, but merely provides that the writ shall not (except in case of rebellion or invasion) be suspended. See R. Fallon, D. Meltzer, & D. Shapiro, Hart & Wechsler’s The Federal Courts and the Federal System 1369 (4th ed. 1996) (“[T]he text [of the Suspension Clause] does not confer a right to habeas relief, but merely sets forth when the ‘Privilege of the Writ’ may be suspended”). Indeed, that was precisely the objection expressed by four of the state ratifying conventions — that the Constitution failed affirmatively to guarantee a right to habeas corpus. See Collings, Habeas Corpus for Convicts — Constitutional Right or Legislative Grace?, 40 Calif. L. Rev. 335, 340, and nn. 39-41 (1952) (citing 1 J. Elliott, Debates on the Federal Constitution 328 (2d ed. 1836) (New York); 3 id., at 658 (Virginia); 4 id., at 243 (North Carolina); 1 id., at 334 (Rhode Island)).

To “suspend” the writ was not to fail to enact it, much less to refuse to accord it particular content. Noah Webster, in his American Dictionary of the English Language, defined it — with patriotic allusion to the constitutional text— as “[t]o cause to cease for a time from operation or effect; as, to suspend the habeas corpus act.” Vol. 2, p. 86 (1828 ed.). See also N. Bailey, An Universal Etymological English Dictionary (1789) (“To Suspend [in Law] signifies a temporal stop of a man’s right”); 2 S. Johnson, A Dictionary of the. English Language 1958 (1773) (“to make to stop for a time”). This was a distinct abuse of majority power, and one that had manifested itself often in the Framers’ experience: temporarily but entirely eliminating the “Privilege of the Writ” for a certain geographic area or areas, or for a certain class *338or classes of individuals. Suspension Acts had been adopted (and many more proposed) both in this country and in England during the late 18th century, see B. Mian, American Habeas Corpus: Law, History, and Politics 109-127 (1984)— including a 7-month suspension by the Massachusetts Assembly during Shay’s Rebellion in 1787, id., at 117. Typical of the genre was the prescription by the Statute of 1794, 34 Geo. 3, c. 54, §2, that “‘[an Act for preventing wrongous imprisonment, and against undue delays in trials], insofar as the same may be construed to relate to the cases of Treason and suspicion of Treason, be suspended [for one year] Mian, supra, at 110. See also 16 Annals of Cong. 44, 402-425 (1852) (recording the debate on a bill, reported to the House of Representatives from the Senate on January 26, 1807, and ultimately rejected, to “suspen[d], for and during the term of three months,” “the privilege of the writ of habeas corpus” for “any person or persons, charged on oath with treason, misprision of treason,” and other specified offenses arising out of the Aaron Burr conspiracy).

In the present case, of course, Congress has not temporarily withheld operation of the writ, but has permanently altered its content. That is, to be sure, an act subject to majoritarian abuse, as is Congress’s framing (or its determination not to frame) a habeas statute in the first place. But that is not the majoritarian abuse against which the Suspension Clause was directed. It is no more irrational to guard against the common and well known “suspension” abuse, without guaranteeing any particular habeas right that enjoys immunity from suspension, than it is, in the Equal Protection Clause, to guard against unequal application of the laws, without guaranteeing any particular law which enjoys that protection. And it is no more acceptable for this Court to write a habeas law, in order that the Suspension Clause might have some effect, than it would be for this Court to write other laws, in order that the Equal Protection Clause might have some effect.

*339The Court cites many cases which it says establish that it is a “serious and difficult constitutional issue,” ante, at 305, whether the Suspension Clause prohibits the elimination of habeas jurisdiction effected by IIRIRA. Every one of those cases, however, pertains not to the meaning of the Suspension Clause, but to the content of the habeas corpus provision of the United States Code, which is quite a different matter. The closest the Court can come is a statement in one of those cases to the effect that the Immigration Act of 1917 “had the effect of precluding judicial intervention in deportation cases except insofar as it was required by the Constitution,” Heikkila, 345 U.(S., at 234-235. That statement (1) was pure dictum, since the Court went on to hold that the judicial review of petitioner’s deportation order was unavailable; (2) does not specify to what extent judicial review was “required by the Constitution,” which could (as far as the Court’s holding was concerned) be zero; and, most important of all, (3) does not refer to the Suspension Clause, so could well have had in mind the due process limitations upon the procedures for determining deportability that our later cases establish, see Part III, infra.

There is, however, another Supreme Court dictum that is unquestionably in point — an unusually authoritative one at that, since it was written by Chief Justice Marshall in 1807. It supports precisely the interpretation of the Suspension Clause I have set forth above. In Ex parte Bollman, 4 Cranch 75, one of the cases arising out of the Burr conspiracy, the issue presented was whether the Supreme Court had the power to issue a writ of habeas corpus for the release of two prisoners held for trial under warrant of the Circuit Court of the District of Columbia. Counsel for the detainees asserted not only statutory authority for issuance of the writ, but inherent power. See id., at 77-93. The Court would have nothing to do with that, whether under Article III or any other provision. While acknowledging an inherent power of the courts “over their own officers, or *340to protect themselves, and their members, from being disturbed in the exercise of their functions,” Marshall says that “the power of taking cognizance of any question between individuals, or between the government and individuals,”

“must be given by written law.

“The inquiry, therefore, on this motion will be, whether by any statute compatible with the constitution of the United States, the power to award a writ of habeas corpus, in such a case as that of Erick Bollman and Samuel Swartwout, has been given to this court.” Id., at 94.

In the ensuing discussion of the Judiciary Act of 1789, the opinion specifically addresses the Suspension Clause — not invoking it as a source of habeas jurisdiction, but to the contrary pointing out that without legislated habeas jurisdiction the Suspension Clause would have no effect.

“It may be worthy of remark, that this act was passed by the first congress of the United States, sitting under a constitution which had declared ‘that the privilege of the writ of habeas corpus should not be suspended, unless when, in cases of rebellion or invasion, the public safety might require it.’

“Acting under the immediate influence of this injunction, they must have felt, with peculiar force, the obligation of providing efficient means by which this great constitutional privilege should receive life and activity; for if the means be not in existence, the privilege itself would be lost, although no law for its suspension should be enacted. Under the impression of this obligation, they give to all the courts the power of awarding writs of habeas corpus” Id., at 95.5

*341There is no more reason for us to believe, than there was for the Marshall Court to believe, that the Suspension Clause means anything other than what it says.

B

Even if one were to assume that the Suspension Clause, despite its text and the Marshall Court’s understanding, guarantees some constitutional minimum of habeas relief, that minimum would assuredly not embrace the rarified right asserted here: the right to judicial compulsion of the exercise of Executive discretion (which may be exercised favorably or unfavorably) regarding a prisoner’s release. If one reads the Suspension Clause as a guarantee of habeas relief, the obvious question presented is: What habeas relief? There are only two alternatives, the first of which is too absurd to be seriously entertained. It could be contended that Congress “suspends” the writ whenever it eliminates any prior ground for the writ that it adopted. Thus, if Congress should ever (in the view of this Court) have authorized immediate habeas corpus — without the need to exhaust administrative remedies — for a person arrested as an illegal alien, Congress would never be able (in the light of sad experience) to revise that disposition. The Suspen*342sion Clause, in other words, would be a one-way ratchet that enshrines in the Constitution every grant of habeas jurisdiction. This is, as I say, too absurd to be contemplated, and I shall contemplate it no further.

The other alternative is that the Suspension Clause guarantees the common-law right of habeas corpus, as it was understood when the Constitution was ratified. There is no doubt whatever that this did not include the right to obtain discretionary release. The Court notes with apparent credulity respondent’s contention “that there is historical evidence of the writ issuing to redress the improper exercise of official discretion,” ante, at 303-304. The only framing-era or earlier cases it alludes to in support of that contention, see ante, at 303, n. 23, referred to ante, at 303-304, establish no such thing. In Ex parte Boggin, 13 East 549, 104 Eng. Rep. 484 (K. B. 1811), the court did not even bother calling for a response from the custodian, where the applicant failed to show that he was statutorily exempt from impressment under any statute then in force. In Chalacombe’s Case, reported in a footnote in Ex parte Boggin, the court did “let the writ go” — i. e., called for a response from the Admiralty to Chalacombe’s petition — even though counsel for the Admiralty had argued that the Admiralty’s general policy of not impressing “seafaring persons of [Chalacombe’s] description” was “a matter of grace and favour, [and not] of right.” But the court never decided that it had authority to grant the relief requested (since the Admiralty promptly discharged Chalacombe of its own accord); in fact, it expressed doubt whether it had that authority. See 13 East, at 550, n. (b), 104 Eng. Rep., at 484, n. (a)2 (Lord Ellenborough, C. J.) (“[Considering it merely as a question of discretion, is it not more fit that this should stand over for the consideration of the Admiralty, to whom the matter ought to be disclosed?”). And in Hollingshead’s Case, 1 Salkeld 351, 91 Eng. Rep. 307 (K. B. 1702), the “warrant of commitment” issued by the “commissioners of bankrupt” was “held naught,” since it au*343thorized the bankrupt’s continued detention by the commissioners until “otherwise discharged by due course of law,” whereas the statute authorized commitment only “till [the bankrupt] submit himself to be examined by the commissioners.” (Emphasis deleted.) There is nothing pertaining to executive discretion here.

All the other framing-era or earlier cases cited in the Court’s opinion — indeed, all the later Supreme Court cases until United States ex rel. Accardi v. Shaughnessy, 347 U. S. 260, in 1954 — provide habeas relief from executive detention only when the custodian had no legal authority to detain. See 3 J. Story, Commentaries on the Constitution of the United States § 1333, p. 206 (1833) (the writ lies to ascertain whether a “sufficient ground of detention appears”). The fact is that, far from forming a traditional basis for issuance of the writ of habeas corpus, the whole “concept of ‘discretion’ was not well developed at common law,” Hafetz, The Untold Story of Noncriminal Habeas Corpus and the 1996 Immigration Acts, 107 Yale L. J. 2509, 2534 (1998), quoted in Brief for Respondent in Calcano-Martinez v. INS, O. T. 2000, No. 00-1011, p. 37. An exhaustive search of cases antedating the Suspension Clause discloses few instances in which courts even discussed the concept of executive discretion; and on the rare occasions when they did, they simply confirmed what seems obvious from the paucity of such discussions — namely, that courts understood executive discretion as lying entirely beyond the judicial ken. See, e. g., Chalacombe’s Case, supra, at 342. That is precisely what one would expect, since even the executive’s evaluation of the facts — a duty that was a good deal more than discretionary — was not subject to review on habeas. Both in this country, until passage of the Habeas Corpus Act of 1867, and in England, the longstanding rule had been that the truth of the custodian’s return could not be controverted. See, e. g., Opinion on the Writ of Habeas Corpus, Wilm. 77, 107, 97 Eng. Rep. 29, 43 (H. L. 1758); Note, Developments in *344the Law — Federal Habeas Corpus, 83 Harv. L. Rev. 1038, 1113-1114, and nn. 9-11 (1970) (quoting Act of Feb. 5, 1867, ch. 28, § 1, 14 Stat. .385); Oaks, Legal History in the High Court — Habeas Corpus, 64 Mich. L. Rev. 451, 453 (1966). And, of course, going beyond inquiry into the legal authority of the executive to detain would have been utterly incompatible with the well-established limitation upon habeas relief for a convicted prisoner: “[0]nce a person had been convicted by a superior court of general jurisdiction, a court disposing of a habeas corpus petition could not go behind the conviction for any purpose other than to verify the formal jurisdiction of the committing court.” Id., at 468, quoted in Swain v. Pressley, 430 U. S. 372, 384-385 (1977) (Burger, C. J., concurring in part and concurring in judgment).

In sum, there is no authority whatever for the proposition that, at the time the Suspension Clause was ratified — or, for that matter, even for a century and a half thereafter — habeas corpus relief was available to compel the Executive’s allegedly wrongful refusal to exercise discretion. The striking proof of that proposition is that when, in 1954, the Warren Court held that the Attorney General’s alleged refusal to exercise his discretion under the Immigration Act of 1917 could be reviewed on habeas, see United States ex rel. Accardi v. Shaughnessy, supra, it did so without citation of any supporting authority, and over the dissent of Justice Jackson, joined by three other Justices, who wrote:

“Of course, it may be thought that it would be better government if even executive acts of grace were subject to judicial review. But the process of the Court seems adapted only to the determination of legal rights, and here the decision is thrusting upon the courts the task of reviewing a discretionary and purely executive function. Habeas corpus, like the currency, can be debased by over-issue quite as certainly as by too niggardly use. We would ... leave the responsibility for suspension or *345execution of this deportation squarely on the Attorney General, where Congress has put it.” Id., at 271.

III

Given the insubstantiality of the due process and Article III arguments against barring judicial review of respondent’s claim (the Court does not even bother to mention them, and the Court of Appeals barely acknowledges them), I will address them only briefly.

The Due Process Clause does not “[r]equir[e] [jjudicial [determination [o]f ” respondent’s claim, Brief for Petitioners in Calcano-Martinez v. INS, O. T. 2000, No. 00-1011, p. 34. Respondent has no legal entitlement to suspension of deportation, no matter how appealing his case. “[T]he Attorney General’s suspension of deportation [is] “an act of grace” which is accorded pursuant to her ‘unfettered discretion,’ Jay v. Boyd, 351 U. S. 345,354 (1956)..., and [can be likened, as Judge Learned Hand observed,] to “a judge’s power to suspend the execution of a sentence, or the President’s to pardon a convict,” 351 U. S., at 354, n. 16 . . . .” INS v. Yueh-Shaio Yang, 519 U. S. 26, 30 (1996). The furthest our cases have gone in imposing due process requirements upon analogous exercises of Executive discretion is the following. (1) We have required “minimal procedural safeguards” for death-penalty clemency proceedings, to prevent them from becoming so capricious as to involve “a state official flipp[ing] a coin to determine whether to grant clemency,” Ohio Adult Parole Authority v. Woodard, 523 U. S. 272, 289 (1998) (O’Connor, J., concurring in part and concurring in judgment). Even assuming that this holding is not part of our “death-is-different” jurisprudence, Shafer v. South Carolina, 532 U. S. 36, 55 (2001) (Scalia, J., dissenting) (citation omitted), respondent here is not complaining about the absence of procedural safeguards; he disagrees with the Attorney General’s judgment on a point of law. (2) We have recognized the existence of a due process liberty interest when *346a State’s statutory parole procedures prescribe that a prisoner “shall” be paroled if certain conditions are satisfied, see Board of Pardons v. Allen, 482 U. S. 369, 370-371, 381 (1987); Greenholtz v. Inmates of Neb. Penal and Correctional Complex, 442 U. S. 1, 12 (1979). There is no such statutory entitlement to suspension of deportation, no matter what the facts. Moreover, in neither Woodard, nor Allen, nor Greenholtz did we intimate that the Due Process Clause conferred jurisdiction of its own force, without benefit of statutory authorization. All three cases were brought under 42 U. S. C. § 1983.

Article III, § l’s investment of the “judicial Power of the United States” in the federal courts does not prevent Congress from committing the adjudication of respondent’s legal claim wholly to “non-Article III federal adjudicative bodies,” Brief for Petitioners in Calcano-Martinez v. INS, O. T. 2000, No. 00-1011, at 38. The notion that Article III requires every Executive determination, on a question of law or of fact, to be subject to judicial review has no support in our jurisprudence. Were it correct, the doctrine of sovereign immunity would not exist, and the APA’s general permission of suits challenging administrative action, see 5 U. S. C. § 702, would have been superfluous. Of its own force, Article III does no more than commit to the courts matters that are “the stuff of the traditional actions at common law tried by the courts at Westminster in 1789,” Northern Pipeline Constr. Co. v. Marathon Pipe Line Co., 458 U. S. 50, 90 (1982) (Rehnquist, J., concurring in judgment) — which (as I have discussed earlier) did not include supervision of discretionary Executive action.

* * *

The Court has created a version of IIRIRA that is not only unrecognizable to its framers (or to anyone who can read) but gives the statutory scheme precisely the opposite of its intended effect, affording criminal aliens more opportu*347nities for delay-inducing judicial review than others have, or even than criminal aliens had prior to the enactment of this legislation. Because § 2241’s exclusion of judicial review is unmistakably clear, and unquestionably constitutional, both this Court and the courts below were without power to entertain respondent’s claims. I would set aside the judgment of the court below and remand with instructions to have the District Court dismiss for want of jurisdiction. I respectfully dissent from the judgment of the Court.