Concurring.
Athough I agree with the majority’s result and analysis with respect to the jurisdictional and equal protection issues in this case, I write separately because of my different approach to the question of the retroactive application of AEDPA § 440(d), Pub.L. No. 104-142, 110 Stat. 1276 (1996).
The immigration-related sections of Title IV of AEDPA contain various provisions expressly providing for their prospective or retrospective application, both as to pre-Act proceedings and pre-Act convictions. These result in a proliferation of sometimes conflicting inferences, making application of statutory construction methods considerably more difficult here than it was in Lindh v. Murphy, 521 U.S. 320, 117 S.Ct. 2059, 138 L.Ed.2d 481 (1997). In light of several relevant AED-PA provisions, I ultimately conclude that the majority is correct in finding ambiguous Congress’s intent with respect to the particular factual situations of the petitioners in these cases, and in proceeding to apply Landgraf v. U.S.I. Film Products, 511 U.S. 244, 114 S.Ct. 1483, 128 L.Ed.2d 229 (1994), to determine whether its presumption against retroactivity applies. My reasons for finding ambiguity, howéver, are narrower than those of the majority. I find the reasoning of our sister circuits in confronting the similar, but legally distinct, issue of the application of § 440(d) to applications for INA § 212(c) relief in proceedings pending as of the effective date of AEDPA to be more problematic for our result today than does the majority. Thus, I wish to explain why I believe our relatively narrow result is not inconsistent with those cases.
The majority opinion concludes that Mayers v. INS, 175 F.3d 1289 (11th Cir.1999), Sandoval v. Reno, 166 F.3d 225 (3d Cir.1999), Henderson v. INS, 157 F.3d 106 (2d Cir.1998), and Goncalves v. Reno, 144 F.3d 110 (1st Cir.1998), “overlook” ostensibly contradictory effective date provisions of other sections of AEDPA Title IV, specifically AEDPA §§ 435 and 441. Maj. Op. at 1149-50. I disagree. Discussing the absence of an explicit effective date provision in § 440, the Third Circuit noted that “several other comparable provisions *1154do have express effective dates,” and then proceeded to discuss § 413, “Denial of Other Relief for Alien Terrorists.” Sandoval, 166 F.3d at 241 (emphasis added); see also Goncalves, 144 F.3d at 128-31. I am of the view that §§ 413 and 421 are much more directly “comparable” to § 440(d) than are §§ 435 and 441. Section 413 deprives certain aliens classified as “alien terrorists” of formerly available discretionary relief. See AEDPA § 413 (amending former 8 U.S.C. § 1253(h), which provided for withholding of deportation based on determination by Attorney General). Section 421 restricts discretionary relief applicable to asylum applicants. See AEDPA § 421. These restrictions on discretionary relief are directly comparable to that of § 440(d), and therefore the negative implication stemming from their explicit effective dates clearly applies to § 440(d).
Sections 435 and 441, on the other hand, are not similarly comparable, as they do not operate to restrict a previously available form of discretionary relief, but rather respectively alter those crimes of moral turpitude serving as grounds for deportation, see § 435, and restrict the availability of collateral attacks on a deportation order, see § 441.
Given the more direct parallels between the purposes of § 440(d) and §§ 413 and 421, I believe that the presence of an explicit directive for prospective application of other, less comparable, provisions of AEDPA, does not necessarily refute the negative implication to be drawn from §§ 413 and 421, at least with respect to the application of § 440(d) to proceedings pending as of AEDPA’s enactment. Only because I conclude that § 440(f) prevents the implication drawn from §§ 413 and 421 from resolving decisively the application of § 440(d) to the situation before us today— post-AEDPA proceedings implicating pre-AEDPA convictions — do I agree that congressional intent is ultimately ambiguous as to that restricted set of cases.1 Thus, this particular ambiguity does not necessarily undermine the reasoning of Sandoval and Goncalves.
It is especially appropriate for courts to resort to legislative history for guidance when the language and structure of a statute are ambiguous. See United States v. Donruss Co., 393 U.S. 297, 303, 89 S.Ct. 501, 21 L.Ed.2d 495 (1969). Therefore, I would not “decline to find that Congress manifested a clear intent one way or the other,” Maj. Op. at 1150, without considering the legislative history arguments against retrospective application of § 440(d). See, e.g., Goncalves, 144 F.3d at 131-33 (analyzing legislative history of § 440(d) and concluding that Congress intended § 440(d) not apply retrospectively to pending applications); Mayers, 175 F.3d 1289, 1303-04 (same); Sandoval, 166 F.3d at 241 (same); Henderson, 157 F.3d at 130 (same). Most significantly, the original Senate language making § 440(d) retroactive as to “cases pending before, on, or after such date of enactment” was removed in conference. See 141 Cong. Rec. S7559 (daily ed. May 25, 1995); see generally Goncalves, 144 F.3d at 131-32 (detailing legislative history). As noted in Sandoval, “[f]ew principles of statutory construction are more compelling than the proposition that Congress does not intend sub silentio to enact statutory language that it has earlier discarded in favor of other language.” 166 F.3d at 241 (quoting INS v. Cardoza-Fonseca, 480 U.S. 421, 442-43, 107 S.Ct. 1207, 94 L.Ed.2d 434 (1987)). In the cases before us, however, this implication is considerably weaker. The ultimately excluded Senate language spoke only to the issue of application of § 440(d) to pending proceedings, and not to the distinct issue of application of § 440(d), in post-Act proceedings, to pre-Act conduct.2
*1155Therefore, I agree with the majority that normal methods of statutory construction compel a finding of ambiguity with respect to the particular situation presented in these eases, but I find this ambiguity results solely from an implication, stemming from AEDPA § 440(f), that § 440(d) may be intended to apply to pre-Act convictions. This subsection, entitled “Effective Date,” provides that (with one exception as to § 440(e)(3)), “[t]he amendments made by subsection (e) shall apply to convictions entered on or after the date of the enactment of this Act.” AEDPA § 440(f). Subsection (e) operates to amend the definition of “aggravated felony” in 8 U.S.C. § 1101(a)(43), expanding the range of offenses covered within that term. Subsection 440(f) is thus an explicit provision making prospective the application of one subsection of § 440, § 440(e), as to convictions. The fact that § 440(f) makes § 440(e) operate only prospectively as applied to convictions may permit an inference that other subsections of the same statutory section are intended to apply retrospectively to convictions. Cf. Lindh, 521 U.S. at 326-27, 117 S.Ct. 2059 (finding that provision for retroactive application in one statutory chapter creates a negative implication for analogous chapter about whose temporal application the statute is silent); cf. De Sousa v. Reno, 30 F.Supp.2d 844, 853 (concluding that the “only way to read section 440(f), as actually drafted, to give it full effect is to conclude that Congress intended to treat past criminal convictions differently in sections 440(d) and 440(e) — i.e., section 440(d) applies to prior convictions, while section 440(e) does not”).3 Because this implication conflicts with those of §§ 413 and 421, I agree with the majority that standard statutory construction does not allow us to derive a clear congressional intent as to the application of § 440(d), in post-Act proceedings, to pre-Act convictions.
Having concluded that legislative intent is ambiguous as to retrospective application for that narrow set of eases involving post-Act proceedings based on pre-Act criminal convictions, I agree with the majority that Lindh and Landgraf thus require us to resort to analysis of whether the provision has retroactive effect so as to invoke the traditional judicial “presumption against retroactivity.” See Landgraf, 511 U.S. at 272, 114 S.Ct. 1483. I further agree with the majority that this limited set of cases — where the law revokes a form of discretionary relief in post-Act proceedings, rather than imposing new consequences on the underlying convictions — does not sufficiently implicate the concerns of “fair notice, reasonable reliance, and settled expectation” to invoke the presumption against retroactivity. Landgraf, 511 U.S. at 270, 114 S.Ct. 1483.
Accordingly, I concur.
. The majority correctly makes clear the cases before us do not implicate the distinct legal situation presented in Mayers, Sandoval, Henderson, and Goncalves that of cases pending at the time of AEDPA's enactment. See Maj. Qp. at 1147 n. 9.
. The additional legislative history argument discussed in Goncalves is also considerably *1155less persuasive outside the context of pending proceedings, and thus does not offer us sufficient certainty to resolve, in this case, the tension that emerges from the structure of the statute itself. See Goncalves, 144 F.3d at 133 (noting that Congress was aware of and did not react to, at the time of amending § 440(d) in September 1996, a BIA decision providing for prospective application to pending cases, but also noting that such subsequent legislative developments are "never determinative in themselves”).
. In light of the contrary implications of §§413 and 421, I am not as certain as the court in De Sousa that § 440(1) permits a conclusion of clear legislative intent, but am convinced it does permit a conclusion of ambiguous intent.