OPINION OF THE COURT
Read, J.The United States Supreme Court held in Padilla v Kentucky (559 US 356 [2010]) that the Sixth Amendment requires crimi*782nal defense counsel to advise their noncitizen clients about the risk of deportation arising from a guilty plea. The Court subsequently held in Chaidez v United States (568 US —, 133 S Ct 1103 [2013]) that Padilla did not apply retroactively in federal collateral review. The issue in this appeal is whether, pursuant to federal or state retroactivity principles, Padilla nonetheless applies retroactively in state court postconviction proceedings. For the reasons that follow, we hold that it does not, and therefore reverse the Appellate Division.
L
On April 20, 1995, defendant Roman Baret was indicted on six counts of third-degree sale of a controlled substance (Penal Law § 220.39) and six counts of third-degree possession of a controlled substance (Penal Law § 220.16). He was also charged, while acting in concert with a codefendant, with an additional two counts of third-degree possession and one count of fourth-degree possession (Penal Law § 220.09). On December 23, 1996, defendant pleaded guilty to one count of third-degree sale in exchange for an indeterminate prison term of 2 to 6 years and a recommendation for shock incarceration. The plea fulfilled his part of a “no-split” offer under which his codefendant would plead guilty and receive probation.
During the plea colloquy, defendant admitted to the sale, acknowledged that a guilty plea was the equivalent of a conviction after trial and that, by pleading guilty, he was giving up his rights to remain silent, seek suppression of evidence and present his own evidence at trial. He acknowledged that he could receive between 873 to 25 years in prison if convicted after trial. Defendant assured the judge that no one had forced him to plead guilty, and the judge warned that he was authorized to sentence defendant to the maximum term of incarceration if he failed to return for sentencing.
Defendant, represented by new counsel, subsequently moved to withdraw his plea; he claimed that he had pleaded guilty because of threats made by his codefendant. The judge denied the motion. Defendant then failed to appear for sentencing, and a bench warrant was issued. More than seven years later, on October 20, 2004, he was returned to court involuntarily. On December 20, 2004, the court (a new judge) sentenced defendant to the originally promised sentence of 2 to 6 years’ incarceration.
*783On appeal, defendant argued that the judge who took the plea abused his discretion when he denied the motion to withdraw without holding a hearing. The Appellate Division disagreed and affirmed, with two Justices dissenting (43 AD3d 648 [1st Dept 2007]). A dissenting Justice granted defendant’s application for leave to appeal to us (2007 NY Slip Op 83234[U] [2007]), and we affirmed (11 NY3d 31 [2008]).
In December 2010, defendant moved to vacate his conviction pursuant to CPL 440.10 on the ground that defense counsel was ineffective for failing to advise him of the immigration consequences of his guilty plea, entered 14 years earlier. He relied on the Supreme Court’s then-recent decision in Padilla. Defendant claimed that he would have rejected the plea and gone to trial if he had known that, by pleading guilty, he became subject to deportation.
In a decision and order dated March 3, 2011, Supreme Court declined to apply Padilla retroactively to defendant’s claim. The court next concluded that defendant had failed to show that his attorney was otherwise ineffective under Strickland v Washington (466 US 668 [1984]) in light of the generous plea bargain he received. Under the two-prong Strickland test, a defendant must show both that his attorney’s conduct fell below established professional norms and that there is a reasonable probability that, but for the error, the proceeding would have turned out differently (id. at 687-688, 694).
On September 1, 2011, a Justice of the Appellate Division granted defendant’s application for leave to appeal the denial of his CPL 440.10 motion (2011 NY Slip Op 82488[U] [2011]). The Appellate Division employed the framework described by the Supreme Court in Teague v Lane (489 US 288 [1989]) and adopted by us in People v Eastman (85 NY2d 265, 275-276 [1995]) to analyze whether Padilla should be applied retroactively (99 AD3d 408, 409 [1st Dept 2012]).
Teague established as a guiding principle that new rules of federal constitutional criminal procedure do not apply retroactively to cases that had become final on direct review before the new rule was announced (see Whorton v Bockting, 549 US 406, 416 [2007] [“Under the Teague framework, an old rule applies both on direct and collateral review, but a new rule is generally *784applicable only to cases that are still on direct review”]).1 Additionally, Teague created a test to differentiate a new rule from an existing or old rule; namely,
“[i]n general, ... a case announces a new rule when it breaks new ground or imposes a new obligation on the States or the Federal Government. To put it differently, a case announces a new rule if the result was not dictated by precedent existing at the time the defendant’s conviction became final” (489 US at 301 [citations omitted]).
Subsequent decisions defined “dictated by precedent” to mean that the result was “apparent to all reasonable jurists” at the time the defendant’s conviction became final (Lambrix v Singletary, 520 US 518, 527-528 [1997]; see also Graham v Collins, 506 US 461, 477 [1993] [the “determinative question is whether reasonable jurists reading the case law that existed in 1984 could have concluded that (the defendant’s) sentencing was not constitutionally infirm”]; Butler v McKellar, 494 US 407, 415 [1990] [a constitutional rule that is “susceptible to debate among reasonable minds” qualifies as a new rule]).
The Teague court also fashioned two exceptions to this general rule. First, “a new rule should be applied retroactively if it places certain kinds of primary, private individual conduct beyond the power of the criminal law-making authority to proscribe” (489 US at 311 [internal quotation marks omitted]). The second exception was “reserved for watershed rules of criminal procedure” (id.), defined generally as “those new procedures [of fundamental fairness] without which the likelihood of an accurate conviction is seriously diminished” (id. at 313).
As the Appellate Division acknowledged, we have long considered deportation to be a collateral consequence of a guilty *785plea (see 99 AD3d at 409, citing People v Ford, 86 NY2d 397, 405 [1995]).2 As a result, prior to Padilla we held in Ford that counsel’s failure to warn a defendant that a guilty plea might lead to removal from the United States did not amount to ineffective assistance of counsel under the Federal or State Constitutions (id. at 405), while inaccurate advice about a guilty plea’s immigration consequences constituted ineffective assistance under the Federal Constitution (see People v McDonald, 1 NY3d 109, 111 [2003]).3 The Appellate Division nonetheless decided in favor of retroactivity on the theory that
“Padilla did not establish a new rule under Teague; rather, it followed from the clearly established principles of the guarantee of effective assistance of counsel under Strickland, and merely clarified the law as it applied to the particular facts. Rather than overrule a clear past precedent, Padilla held that Strickland applies to advice concerning deportation, whether it be incorrect advice or no advice at all” (99 AD3d at 409 [internal quotation marks and citations omitted]).
Thus, the court held that Padilla was to be retroactively applied to pleas taken after Congress made “significant changes in immigration law” in 1996,4 and “expressed] no opinion on [its] *786applicability ... to pleas taken before” that year (id.). The court remanded for a hearing to determine “what advice, if any, counsel gave defendant regarding the immigration consequences of his plea, and, assuming the advice was constitutionally deficient, whether there is a reasonable probability that but for this deficiency, defendant would have gone to trial” (id. at 410).
The Appellate Division issued its decision and order on October 2, 2012. The Supreme Court handed down Chaidez on February 20, 2013. And on June 5, 2013, a Judge of this Court granted the People leave to appeal (21 NY3d 1002 [2013]).
II.
Chaidez
Roselva Chaidez entered the United States from Mexico in 1971 and became a lawful permanent resident in 1977. Chaidez was indicted on three counts of mail fraud because of her involvement in a staged automobile accident insurance scam. On advice of counsel, she pleaded guilty to two counts and was sentenced to four years of probation and ordered to pay restitution. Her conviction became final in 2004. Under federal immigration law, the crimes to which Chaidez pleaded are “aggravated felonies,” and conviction of an aggravated felony is virtually certain to result in deportation (see 8 USC § 1227 [a] [2] [A] [iii]; id. § 1101 [a] [43] [M] [i] [an offense that involves fraud or deceit in which the loss to the victim exceeds $10,000 is an aggravated felony]). According to Chaidez, though, her attorney never advised her of this.
Chaidez later applied for naturalization, which brought her convictions to the attention of federal immigration authorities. They promptly initiated removal proceedings against her. She then sought a writ of coram nobis in federal district court to overturn her 2004 conviction, claiming ineffective assistance of counsel in connection with her decision to plead guilty. Chaidez maintained that she never would have pleaded guilty if her attorney had made her aware of the immigration consequences of such a plea. While the writ was pending, the Supreme Court handed down its decision in Padilla.
Applying Teague, the federal district court held that Padilla did not announce a new rule, but merely applied Strickland to *787new facts (United States v Chaidez, 730 F Supp 2d 896, 904 [ND Ill 2010]). After holding a hearing on Chaidez’s corara nobis petition, the court determined that she was entitled to relief for ineffective assistance pursuant to Strickland, granted the petition and vacated her conviction (2010 WL 3979664, 2010 US Dist LEXIS 107877 [ND Ill, Oct. 6, 2010, No. 03 CR 636-6]). On appeal, the United States Court of Appeals for the Seventh Circuit reversed, concluding that Padilla had announced a new rule within the meaning of Teague (Chaidez v United States, 655 F3d 684, 686 [2011]).5 The Supreme Court agreed and therefore affirmed, holding that, under the Teague framework, Padilla does not apply in collateral challenges to final convictions.
Writing for a six-Justice majority,6 Justice Kagan stated as a basic premise that “ ‘a rule designed for the specific purpose of evaluating a myriad of factual contexts . . . will . . . infrequent[ly] . . . yield[ ] a result so novel that it forges a new rule . . . not dictated by precedent’ ” within the meaning of Teague (Chaidez, 568 US at —, 133 S Ct at 1107, quoting Wright v West, 505 US 277, 309 [1992, Kennedy, J., concurring in judgment]). As a result, “garden-variety applications” of the Strickland test “do not produce new rules” (568 US at —, 133 S Ct at 1107). But, she continued, Padilla did not “merely [make] clear that a lawyer who neglects to inform a client about the risk of deportation is professionally incompetent” (id. at 1108). Rather, she explained,
“Padilla did something more. Before deciding if failing to provide such advice fell below an objective standard of reasonableness, Padilla considered a *788threshold question: Was advice about deportation categorically removed from the scope of the Sixth Amendment right to counsel because it involved only a collateral consequence of a conviction, rather than a component of the criminal sentence? In other words, prior to asking how the Strickland test applied (‘Did this attorney act unreasonably?’), Padilla asked whether the Strickland test applied (‘Should we even evaluate if this attorney acted unreasonably?’). And [because] . . . that preliminary question about Strickland's ambit came to the Padilla Court unsettled[,] . . . the Court’s answer (‘Yes, Strickland governs here’) required a new rule” (id. [internal quotation marks and citation omitted]).
Justice Kagan then considered as “relevant background” the Court’s decision in Hill v Lockhart (474 US 52 [1985]), which “explicitly left open whether advice concerning a collateral consequence must satisfy Sixth Amendment requirements” (568 US at —, 133 S Ct at 1108). As a result, she observed, resolution of this issue fell to the lower federal and state courts, and all 10 of the federal appellate courts to consider the issue and 30 of the 32 state appellate courts (and here, Justice Kagan cited our decision in Ford) decided that “the Sixth Amendment does not require attorneys to inform their clients of a conviction’s collateral consequences, including deportation” (id. at 1109). This “imbalance,” she remarked, even caused the authors of “the principal scholarly article on the subject to call the exclusion of advice about collateral consequences from the Sixth Amendment’s scope one of ‘the most widely recognized rules of American law’ ” (id., quoting Chin & Holmes, Effective Assistance of Counsel and the Consequences of Guilty Pleas, 87 Cornell L Rev 697, 706 [2002]).
In sum, Justice Kagan concluded, the lower federal and state appellate courts “almost uniformly insisted on what Padilla called the 'categorica[l] remov[al]’ of advice about a conviction’s non-criminal consequences — including deportation — from the Sixth Amendment’s scope”; Padilla was the first case to “reject[ ] that categorical approach — and so [make] the Strickland test operative — when a criminal lawyer gives (or fails to give) advice about immigration consequences”; therefore, Padilla announced a new rule within the meaning of Teague (568 US at —, 133 S Ct at 1110-1111). Indeed, she added, “[i]f that *789does not count as ‘breaking] new ground’ or ‘impos[ing] a new obligation,’ we are hard pressed to know what would” (id. at —, at 1110, quoting Teague, 489 US at 301). Accordingly, the Court affirmed the Seventh Circuit, holding that Padilla constituted a new rule and, as such, would not apply in collateral challenges in federal courts to final convictions.7
III.
Defendant does not regard Chaidez as creating an insuperable obstacle to an affirmance. He asks us to hold that Padilla applies in collateral challenges to convictions in New York courts occurring between 1996, when Congress “severely tightened” immigration laws, and March 31, 2010, when Padilla was handed down, and offers three theories or arguments to support his request: (1) Padilla is a watershed rule of federal constitutional criminal procedure within the meaning of Teague and/or Eastman; or, in light of Danforth we should either (2) interpret Teague more broadly than did the Supreme Court in Chaidez and hold that Padilla is simply an application of Strickland, or (3) apply the three-factor test in People v Pepper (53 NY2d 213 [1981], cert denied sub nom. New York v Utter, 454 US 1162 [1982] ) to make Padilla retroactive on collateral review. Having already discussed Padilla, Strickland, Teague and Chaidez in relevant detail, we now turn to Danforth, Pepper and Eastman.
Danforth
In 1996, a Minnesota jury convicted Danforth of a sex offense involving a six-year-old boy. The victim was ruled incompetent to testify at trial, but the jury saw and heard a videotaped *790interview of him. Danforth unsuccessfully urged on appeal that the tape’s admission violated his right to confrontation secured by the Sixth Amendment. His conviction became final in 1998. After the Supreme Court handed down its decision in Crawford v Washington (541 US 36 [2004]), though, Danforth filed a state postconviction petition. He argued that he was entitled to a new trial because the admission of the taped interview violated the rule announced in Crawford (i.e., that defendants have a right under the Sixth Amendment’s Confrontation Clause to cross-examine testimonial witnesses). Assessing Danforth’s claim in light of Teague, the postconviction and intermediate appeals courts both ruled that Crawford did not apply retroactively.
On appeal to the Minnesota Supreme Court, Danforth for the first time argued that Minnesota courts were free to apply retroactivity more broadly than Teague, and that he was entitled to the benefit of Crawford under state retroactivity principles. He took the position that Teague “dictate[d] the limits of retroactive application of new rules only in federal habeas corpus proceedings and [did] not limit the retroactive application of new rules in state postconviction proceedings” (Danforth v State, 718 NW2d 451, 456 [Minn 2006]). The court chose to address this issue in the interests of justice, but rejected Danforth’s assertions that “state courts are free to give a Supreme Court decision of federal constitutional criminal procedure broader retroactive application than that given by the Supreme Court” (id.). Applying Teague, the court then concluded that Crawford was a new rule which did not fall within the exception for watershed rules and therefore did not apply retroactively.
In explaining the decision to grant certiorari in Danforth, the Supreme Court noted that although the Minnesota Supreme Court “correctly concluded that federal law does not require state courts to apply the holding in Crawford to cases that were final when that case was decided” (Danforth, 552 US at 268-269 [discussing its holding in Whorton (549 US at 409), that Crawford is not a watershed rule]), a question remained about whether Teague “prohibited] [state courts] from doing so” (Danforth, 552 US at 268-269). The Court held that Teague did not.8
In so deciding, the Court first summarized the Teague rule as follows:
*791“New constitutional rules announced by this Court that place certain kinds of primary individual conduct beyond the power of the States to proscribe, as well as ‘watershed’ rules of criminal procedure, must be applied in all future trials, all cases pending on direct review, and all federal habeas corpus proceedings. All other new rules of criminal procedure must be applied in future trials and in cases pending on direct review, but may not provide the basis for a federal collateral attack on a state-court conviction” (id. at 266 [emphasis added]).
Consistent with this formulation, Danforth’s holding derived generally from the notion that Teague was “an exercise of [the] Court’s power to interpret the federal habeas statute” (id. at 278); “was meant to apply only to federal courts considering habeas corpus petitions challenging state-court criminal convictions”; and was “justified ... in part by reference to comity and respect for the finality of state convictions [,] . . . considerations . . . unique to federal habeas review of state convictions” (id. at 279).9 As a consequence, the Court concluded,
“[i]t is thus abundantly clear that the Teague rule of nonretroactivity was fashioned to achieve the goals of federal habeas while minimizing federal intrusion into state criminal proceedings. It was intended to limit the authority of federal courts to overturn state convictions — not to limit a state court’s authority to grant relief for violations of new rules of constitutional law when reviewing its own State’s convictions” (id. at 280-281).10
On remand, the Minnesota Supreme Court elected to retain the Teague standard, principally so as not to “ ‘undermine[ ] *792the principle of finality which is essential to the operation of our criminal justice system’ ” (Danforth v State, 761 NW2d 493, 498 [Minn 2009], quoting Teague, 489 US at 309), and affirmed Danforth’s conviction. The court indicated, however, that it would not necessarily be bound in the future by the Supreme Court’s determination as to whether a new rule was “critical to fundamental fairness” (id. at 500).
Pepper
In People v Samuels (49 NY2d 218 [1980]), we held that once an accusatory instrument has been filed, a defendant cannot waive his constitutional right to counsel in counsel’s absence. The issue for us in Pepper was the “retrospective effect, if any, . . . to be accorded” Samuels in three appeals (Pepper, 53 NY2d at 217). In two of them (Pepper and People v Utter), Samuels was decided while the defendants’ appeals were pending in the Appellate Division. That court reversed the judgments of conviction and sentences and remanded the cases for further proceedings because incriminating statements had been elicited in violation of Samuels. In the third case (People v Torres), Samuels was handed down about a year and a half after a Judge of this Court denied the defendant’s application for leave to appeal, which was subsequently granted upon the defendant’s motion for reconsideration.
Seeking a balance between “full retroactive application (permitting a collateral attack on a conviction no longer in normal appellate channels)” and “limit[ing] relief to prospective police conduct or trials” (53 NY2d at 220), the Pepper court adopted the three-part test set out by the Supreme Court in Linkletter v Walker (381 US 618, 629 [1965]) and recapitulated in Stovall v Denno (388 US 293, 297 [1967]) and Desist (394 US at 249).11 The Pepper test requires us to
*793“weigh three factors to determine whether a new precedent operates retroactively: the purpose to be served by the new standard; the extent of the reliance by law enforcement authorities on the old standard; and the effect on the administration of justice of a retroactive application of the new standard. The second and third factors are, however, only given substantial weight ‘when the answer to the retroactivity question is not to be found in the purpose of the new rule itself. Thus, where otherwise there could be a complete miscarriage of justice, current constitutional standards that go to the heart of a reliable determination of guilt or innocence have been substituted for those in effect at the time of trial’ ” (Policano v Herbert, 7 NY3d 588, 603 [2006] [citation omitted], quoting Pepper, 53 NY2d at 220-221).
Observing that we had previously considered “a defendant’s right to counsel in pretrial encounters [to] fall[ ] within a middle ground” such that retroactivity had previously “been limited to those [cases] still on direct review at the time the change in law occurred,” we saw “no reason to depart from that course” (Pepper, 53 NY2d at 221).
In weighing the three factors, we commented that “uncounseled station house interrogations [do not] necessarily go to the ultimate issue of guilt or innocence, [but] they are not insignificant events”; that “the relatively early terminal point of retroactivity limited to cases young enough to still be on direct appeal militates against the likelihood of prejudicial mildewing of witnesses’ memories”; and that the People could not credibly plead reliance since the holding in Samuels “had been more than foreshadowed” by earlier rulings (id. at 221-222).
Thus, we concluded that the Appellate Division had properly applied Samuels in Pepper’s and Utter’s appeals. Torres, however, was not entitled to benefit from Samuels because
“[t]he normal appellate process came to an end for him upon the original unconditional denial of his application for leave to appeal to this court. Thereafter, only complete retroactivity could make Samuels applicable. And that, impermissibly, would *794mean that every defendant to whose case it was relevant, no matter how remote in time and merit, would become its beneficiary. Absent manifest injustice, no exception was available to Torres” (id. at 222 [citation and internal quotation marks omitted]).
Eastman
The defendant in Eastman moved pursuant to CPL 440.10 to vacate his judgment of conviction and sentence for murder and a weapon offense on the ground that he had been convicted in violation of federal constitutional law, as announced in the Supreme Court’s then-recent decision in Cruz v New York (481 US 186 [1987]).12 The nontestifying codefendant’s redacted confession had been introduced into evidence against Eastman at trial. He argued that Cruz should be applied retroactively in his case. Supreme Court denied the motion, and the Appellate Division affirmed on the basis of harmless error.
We began our analysis with the observations that “the Supreme Court’s holding in Cruz marks a break from both Federal and State law precedents,” and “[b]ecause Cruz fundamentally alters the Federal constitutional landscape, the principles of retroactivity developed by the Supreme Court in construing Federal constitutional law govern the disposition of this case” (Eastman, 85 NY2d at 274-275 [emphasis added]). We then proceeded to evaluate Eastman’s motion in light of Teague.
We concluded first that Cruz stated a new rule within the meaning of Teague as it “unquestionably departs from established precedent, and implicates a bedrock procedural element— the Sixth Amendment right of confrontation” (id. at 276). We then looked at whether Cruz fell within Teague’s exception to non-retroactivity for a watershed rule, and concluded that it did. We reasoned that
“[a]s the rule announced in Cruz is central to an accurate determination of guilt or innocence, the Supreme Court determined that the admission of *795the codefendant’s inculpatory confession against the defendant undermined the fundamental fairness of the trial, where, as here, there was no opportunity for cross-examination to test the reliability of the codefendant’s confession. Therefore, we conclude that retroactive application of Cruz is constitutionally commanded on collateral review of a conviction” (id. [citation omitted]).
Deciding further that the Cruz error was not harmless beyond a reasonable doubt, we reversed the Appellate Division, granted Eastman’s CPL 440.10 motion, vacated his convictions and ordered a new trial.
IV.
Is Padilla a Watershed Rule of Federal Constitutional Criminal Procedure within the Meaning of Teague and/or Eastman ?
Defendant emphasizes that Chaidez argued solely that Padilla was not a new rule; she did not contend, alternatively, that even if the Supreme Court disagreed (as it did), Padilla would qualify for the watershed exception to the non-retroactivity of new rules. Chaidez’s neglect to pursue this tack is not surprising, as the Supreme Court has analogized a watershed rule of federal constitutional criminal procedure to the right to counsel established by Gideon v Wainwright (372 US 335 [1963]; see Teague, 489 US at 311-312; see also Whorton, 549 US at 419 [“we have repeatedly referred (to Gideon) in discussing the meaning of the Teague exception” for watershed rules]). In short, “[t]hat a new procedural rule is ‘fundamental’ in some abstract sense is not enough; the rule must be one ‘without which the likelihood of an accurate conviction is seriously diminished’ ” (Schriro v Summerlin, 542 US 348, 352 [2004], quoting Teague, 489 US at 313).
As the Supreme Court has more recently recapitulated these requirements: “First, the rule must be necessary to prevent an impermissibly large risk of an inaccurate conviction. Second, the rule must alter our understanding of the bedrock procedural elements essential to the fairness of a proceeding” (Whorton, 549 US at 418 [internal quotation marks and citations omitted]). The Court has made clear that, for purposes of this second requirement, it is insufficient that a new rule of federal constitutional criminal procedure is “based on a ‘bedrock’ right” (id. at 420-421). “Instead, ... a new rule must itself constitute *796a previously unrecognized bedrock procedural element that is essential to the fairness of a proceeding”; further, in evaluating whether this requirement has been fulfilled, the Court “again . . . looked to the example of Gideon, and . . . [has] not hesitated to hold that less sweeping and fundamental rules do not qualify” (id. at 421 [internal quotation marks and citation omitted]).
Finally, in Teague, and repeatedly since, the Supreme Court has emphasized that additional procedural rules of such profound importance are “unlikely ... to emerge” (Teague, 489 US at 313; accord Graham v Collins, 506 US 461, 478 [1993]; Sawyer v Smith, 497 US 227, 243 [1990]; Butler, 494 US at 416; Tyler v Cain, 533 US 656, 666 n 7 [2001]; Schriro, 542 US at 352). And no watershed rule has, in fact, ever materialized (see United States v Mandanici, 205 F3d 519, 529 [2d Cir 2000] [noting that “(b)eginning with the rule at issue in Teague, the Court has measured at least eleven new rules, or proposed new rules, of criminal procedure against the criteria for the (watershed) exception and, in every case, has refused to apply the rule at issue retroactively”]; see also Schriro 542 US 348 [2004], supra [rejecting retroactivity for Ring v Arizona (536 US 584 [2002])]; Beard v Banks, 542 US 406, 417-420 [2004] [rejecting retroactivity for Mills v Maryland (486 US 367 [1988])]; Whorton, 549 US at 419 [rejecting retroactivity for Crawford]).
Defendant argues that Padilla fulfills both requirements as a matter of federal and state law. As to the first requirement, he focuses on the notion that noncitizen defendants, particularly those charged with drug offenses, were often “lured into . . . extremely lenient dispositions without knowing the truth — that their convictions will lead to deportation with no remedy, lifetime bars to a green card or citizenship,” and often, with a bit of foresight, their pleas might have been recast to avoid untoward immigration consequences. That a plea bargain may turn out to be far less advantageous than a defendant anticipated, however, does not pose “an impermissibly large risk of an inaccurate conviction” (Whorton, 549 US at 418 [emphasis added]).
As for the second requirement, defendant urges that Padilla does, in fact, constitute a bedrock principle akin to Gideon because in the latter, “the defendant was given the right to free counsel and now in Padilla, defendants are given the right to free immigration counsel on their criminal case.” But Padilla *797does no such thing. The Padilla Court recognized that immigration law was “complex” and often “unclear or uncertain,” and that “[s]ome members of the bar who represent clients facing criminal charges, in either state or federal court or both, may not be well versed in it” (Padilla, 559 US at 369). As a result, Padilla imposes a relatively modest duty on counsel; specifically,
“[w]hen the law is not succinct and straightforward ... a criminal defense attorney need do no more than advise a noncitizen client that pending criminal charges may carry a risk of adverse immigration consequences. But when the deportation consequence is truly clear, . . . the duty to give correct advice is equally clear” (id. [emphases added and footnote omitted]).
This is a far cry from “the right to free immigration counsel.” And even if Padilla did, in fact, give defendants such a right, immigration advice is not critical to an accurate determination of guilt or innocence.
Finally, defendant argues that Eastman only calls for a new rule to “implicate[ ] a bedrock procedural element,” such as, in that case, the Sixth Amendment right to confrontation (Eastman, 85 NY2d at 276). At the time we adopted the Teague framework in Eastman, we clearly assumed — just like the Minnesota Supreme Court — that we were obligated to determine the retroactivity of federal constitutional criminal procedures according to federal, rather than state, law. We said as much (i.e., “the principles of retroactivity developed by the Supreme Court in construing Federal constitutional law govern the disposition of this case”) (id. at 275). If we had thought otherwise, we would have decided the retroactivity issue in Eastman using our already well-established three-part Pepper test. And if we had nonetheless intended to interpret Teague more broadly than the Supreme Court, we presumably would have said so rather than merely depend on the word “implicate” to convey this idea. Additionally, unlike the situation here, the Supreme Court had not already decided (and, in fact, never decided) whether the particular procedural rule at issue in Eastman (i.e., Cruz) applied retroactively in collateral proceedings. Finally, the very narrow way in which the Supreme Court has interpreted the watershed exception is much clearer from the vantage point of 2014 than it would have been when we decided Eastman in 1995, only six years after Teague.
Taking all these considerations into account, we have no doubt that the Supreme Court — if the issue had ever reached *798it — would have disagreed with our assessment that, pursuant to Teague, Cruz was a watershed rule. After all, the Court has decided that Crawford, a considerably more far-reaching expansion of the Sixth Amendment right to confrontation, is not a watershed rule. Equally, given the Supreme Court’s now extensive precedent, we are confident that Padilla does not qualify for Teague’s watershed exception. We understand that Danforth frees us to interpret Teague more broadly than the Supreme Court when addressing retroactivity in the context of state collateral review. The reasons proffered by defendant, however, certainly do not convince us that we should do so in this case. In particular, if every federal constitutional criminal procedure based on a bedrock principle of federal constitutional criminal law is a watershed rule, then the exception will swallow the rule and any semblance of finality will vanish. Moreover, we cannot say, as we did about Cruz, that Padilla is “central to an accurate determination of guilt or innocence,” and safeguards “the fundamental fairness of [a] trial” (Eastman, 85 NY2d at 276).
Should we Interpret Teague More Broadly than Did the Supreme Court in Chaidez and Hold That Padilla is Simply an Application of Strickland?
Defendant encourages us to adopt the reasoning of the two-Justice dissent in Chaidez.13 We consider the majority’s logic to *799be far more compelling, though. The majority determined that Padilla was a new rule and not just an application of the two-prong Strickland test for ineffective assistance of counsel to a new set of facts. In so doing, Justice Kagan emphasized that the lower federal and state courts had almost uniformly excluded advice about the immigration consequences of a guilty plea from the scope of the Sixth Amendment right to effective assistance of counsel.
This was certainly true in New York. In 1995 we held in Ford that defense counsel were not “under a duty to warn defendants of the possible deportation consequences before entering a guilty plea”; and, stated another way, that “the failure of counsel to warn defendant of the possibility of deportation [did not] constitute ineffective assistance of counsel” (Ford, 86 NY2d at 401, 404). Thus, Padilla flatly contradicted and supplanted Ford’s holding as to whether defense attorneys were obligated to advise their noncitizen clients about the immigration consequences of a guilty plea. We agree with Justice Kagan that “[i]f [this] does not count as ‘breaking] new ground’ or ‘impos[ing] a new obligation,’ we are hard pressed to know what would” (Chaidez, 568 US at —, 133 S Ct at 1110, quoting Teague, 489 US at 301). Thus, Padilla created a new rule of federal constitutional criminal procedure in New York which, consistent with Teague and Eastman, does not apply retroactively in CPL 440.10 proceedings.
Does the Three-Part Test in Pepper Make Padilla Retroactive on Collateral Review?
The Pepper factors disfavor retroactivity, too. Under the first and most important factor, “current constitutional standards that go to the heart of a reliable determination of guilt or innocence” will be applied retroactively, but “decisions which *800are only collateral to or relatively far removed from the fact-finding process at trial” apply prospectively only (Pepper, 53 NY2d at 221). Padilla has nothing to do with a reliable determination of guilt or innocence; rather, Padilla assures that noncitizen defendants appreciate the immigration risks that inhere in guilty pleas to crimes they acknowledge during the plea allocution to having committed.
As to the remaining two Pepper factors — i.e., reliance on the old standard and the effect of retroactivity on the administration of justice — here again our decision in Ford looms large. As one defense attorney put it in an affirmation in support of a CPL 440.10 application, he knew that his client was a Mexican national, but “it was [his] practice not to dispense any advice regarding the immigration consequences of entering into a negotiated plea because prevailing Court of Appeals precedent did not require it” (Verdejo, 109 AD3d at 140). Many defense attorneys no doubt warned their noncitizen clients about the immigration consequences of a guilty plea. But in light of Ford (and McDonald), many others were likely reluctant to volunteer advice in an often complex area of law outside their legal specialty. And those who did would have had no reason to create or maintain records to memorialize conversations with their clients on this topic. Courts would have to make factual findings based on off-the-record conversations years after a plea, based on faded memories. Here, for example, defendant entered into the plea he now seeks to challenge in December 1996, nearly two decades ago. Such considerations, coupled with the sheer volume of prosecutions disposed of by guilty plea (see Missouri v Frye, 566 US —, —, 132 S Ct 1399, 1407 [2012] [“(N)inety-four percent of state convictions are the result of guilty pleas”]), weigh heavily against retroactive application of Padilla.
Accordingly, the Appellate Division’s order should be reversed and the Supreme Court order reinstated.
. Although the lead opinion in Teague attracted only a four-Justice plurality, a majority of the Court soon adopted its retroactivity analysis (see Penry v Lynaugh, 492 US 302, 313 [1989] [adopting Teague analysis as majority approach], overruled on other grounds by Atkins v Virginia, 536 US 304 [2002]). Chief Justice Roberts has called Teague the Supreme Court’s “leading modern precedent on retroactivity” (Danforth v Minnesota, 552 US 264, 292 [2008, Roberts, C.J., dissenting]). Teague descends from the approach to retroactivity advocated by Justice Harlan in Desist v United States (394 US 244, 256-269 [1969, Harlan, J., dissenting]) and Mackey v United States (401 US 667, 675-702 [1971, Harlan, J., concurring in judgments in part and dissenting in part]) (see Teague, 489 US at 303-313; Danforth, 552 US at 273-274, 297-300).
. This is still the case (see People v Peque, 22 NY3d 168, 176, 192 and n 6, 204 [2013]). The Supreme Court, for its part, has “never attempted to delineate the world of ‘collateral consequences’ ” (Chaidez, 568 US at — n 5, 133 S Ct at 1108 n 5, citing Padilla, 559 US at 364 and n 8; see also id. at 365-366 [commenting that whether the distinction between direct and collateral consequences is “appropriate” to “define the scope of constitutionally ‘reasonable professional assistance’ required under Strickland ... is a question (the Court) need not consider . . . because of the unique nature of deportation,” which was “uniquely difficult to classify as either a direct or collateral consequence” of a criminal conviction]).
. The defendant in McDonald relied solely on federal constitutional law (1 NY3d at 114). In Padilla, the Supreme Court concluded “there [was] no relevant difference between an act of commission and an act of omission” with respect to counsel’s duty to advise a defendant about the risk of deportation (Padilla, 559 US at 370 [internal quotation marks omitted]).
. In 1996, Congress enacted the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) (Pub L 104-132, 110 US Stat 1214 [enacted Apr. 24, 1996]), and the Illegal Immigration Reform and Immigrant Responsibility Act of 1996 (IIRIRA) (Pub L 104-208, 110 US Stat 3009 [enacted Sept. 30, 1996; eff Apr. 1, 1997]). The Padilla Court referred to the changes to immigration law wrought by AEDPA and IIRIRA as “confirming its] view that, as a *786matter of federal law, deportation is an integral part — indeed, sometimes the most important part — of the penalty that may be imposed on noncitizen defendants who plead guilty to specified crimes” (see Padilla, 559 US at 364 [footnote omitted]).
. The parties in Chaidez agreed that neither of the two Teague exceptions to non-retroactivity applied (see Chaidez, 655 F3d at 688). Accordingly, Chaidez did not argue in the Supreme Court that the Teague watershed exception was relevant (see Chaidez, 568 US at — n 3, 133 S Ct at 1107 n 3).
. Justice Thomas, who dissented in Padilla, still considered that case to be “wrongly decided” and so, in his view, the majority’s Teague analysis was “unnecessary” (568 US at —, 133 S Ct at 1114 [Thomas, J., concurring in judgment only]). Justice Sotomayor dissented in a writing joined by Justice Ginsburg. As Justice Sotomayor summarized her analysis,
“Padilla did nothing more than apply the existing rule oí [Strickland] in a new setting, the same way the Court has done repeatedly in the past: by surveying the relevant professional norms and concluding that they unequivocally required attorneys to provide advice about the immigration consequences of a guilty plea” (id. [Sotomayor, J., dissenting]).
. After Chaidez was decided, the Appellate Division abandoned the reasoning that it had espoused in this case on the ground that “ ‘the principles of retroactivity developed by the Supreme Court in construing Federal constitutional law govern the disposition’ ” of a New York case involving a new rule of federal constitutional criminal procedure (People v Verdejo, 109 AD3d 138, 141 [1st Dept 2013], quoting Eastman, 85 NY2d at 274-275), and the Padilla rule, which “merely prescribes] a duty imposed on counsel,” does not fall within Teague’s exception to non-retroactivity for watershed rules (109 AD3d at 141; see also People v Bent, 108 AD3d 882, 883 [3d Dept 2013] [“Notwithstanding our prior holding that (Padilla) applies retroactively, the United States Supreme Court has declared that Padilla is a ‘new rule,’ which is not retroactive”; therefore, defendant’s CPL 440.10 motion to vacate his judgment of conviction, which was final prior to Padilla, was properly denied] [citation omitted]; People v Andrews, 108 AD3d 727 [2d Dept 2013] [declining to give broader retroactive effect to Padilla than required under Teague]; People v Soodoo, 109 AD3d 1014 [2d Dept 2013] [same]; People v Alvarez, 111 AD3d 843 [2d Dept 2013] [same]).
. The Danforth Court expressed no opinion, however, as to whether states are required to retroactively apply any new rules that the federal courts may determine to be watershed rules (see Danforth, 552 US at 269 n 4).
. Of course, whether or not the Teague framework was originally shaped by considerations peculiar to federal habeas review of state court convictions, in Chaidez the Supreme Court applied Teague in a proceeding where federal coram nobis relief was sought (see Chaidez, 568 US at — n 1, 133 S Ct at 1106 n 1; see also Danforth, 552 US at 269 n 4 [expressing no opinion as to whether Teague governs cases brought under 28 USC § 2255 to attack federal criminal sentences]; id. at 281 n 16 [noting that lower federal courts have applied Teague to motions brought pursuant to 28 USC § 2255]).
. In his dissenting opinion, which Justice Kennedy joined, Chief Justice Roberts took the position that “the retroactivity of new federal rules is a question of federal law binding on States” (552 US at 293). He therefore concluded that because Danforth’s conviction became final before the Crawford decision and the Court held in Whorton that Crawford did not apply in collateral review, “[t]hat should be the end of the matter” (id. at 311).
. As Chief Justice Roberts pointed out in his dissent in Danforth, the Supreme Court walked away from Linkletter in two stages. First, in Griffith v Kentucky (479 US 314 [1987]), decided more than five years after we handed down Pepper, the Court adopted Justice Harlan’s views, as expressed in his separate writings in Desist and Mackey, with regard to the retroactivity of a new rule in cases still on direct appeal. Thus, the Court in Griffith “held that ‘a new rule for the conduct of criminal prosecutions is to be applied retroactively to all cases, state or federal, pending on direct review or not yet final’ ” (Danforth, 552 US at 297, quoting Griffith at 328). Then in Teague, the Court “completed the project of conforming [its] view on the retroactivity *793of new rules of criminal procedure to those of Justice Harlan” (Danforth, 552 US at 299; see also n 1, supra).
. Reversing our decision in People v Cruz (66 NY2d 61 [1985]), the Supreme Court held in Cruz v New York that “where a nontestifying codefendant’s confession incriminating the defendant is not directly admissible against the defendant, . . . the Confrontation Clause bars its admission at their joint trial, even if the jury is instructed not to consider it against the defendant, and even if the defendant’s own confession is admitted against him” (Cruz, 481 US at 193).
. Defendant points out that the Supreme Judicial Court of Massachusetts concluded post-Chaidez that “as a matter of Massachusetts law and consistent with [its] authority as provided in [Danforth], . . . the Sixth Amendment right enunciated in Padilla was not a ‘new’ rule and, consequently, defendants whose State law convictions were final after April 1, 1997, may attack their convictions collaterally on Padilla grounds” (Commonwealth v Sylvain, 466 Mass 422, 423-424, 995 NE2d 760, 762 [2013]). There exist two key differences between Massachusetts and New York. First, although the lower courts in Massachusetts seem to have long held that a defendant need not be informed of the collateral consequences of a plea, including the risk of deportation (see Commonwealth v Fraire, 55 Mass App 916, 917, 774 NE2d 677, 678 [2002] [cited in Chaidez, 568 US at — n 8, 133 S Ct at 1109 n 8]), the Supreme Judicial Court never adopted and endorsed this position, as we did in Ford. Second, prior to Chaidez the Supreme Judicial Court held in Commonwealth v Clarke (460 Mass 30, 44, 949 NE2d 892, 903-904 [2011]) that Padilla was not a new rule within the meaning of Teague and therefore applied retroactively to cases on collateral review. Sylvain simply affirmed the court’s previous decision in Clarke.
Defendant has also brought to our attention the recent decision of the New Mexico Supreme Court in Ramirez v State (333 P3d 240 [NM, June 19, 2014]), *799which declined to follow Chaidez. But as Justice Kagan pointed out, New Mexico is one of only two states whose highest court had held, prior to Padilla, that “an attorney could violate the Sixth Amendment by failing to inform a client about deportation risks or other collateral consequences of a guilty plea” (Chaidez, 568 US at — and n 9, 133 S Ct at 1109 and n 9, citing State v Paredez, 2004-NMSC-036, ¶¶ 17-19, 136 NM 533, 539, 101 P3d 799, 805 [2004]). The Ramirez court applied its Paredez precedent retroactively to 1990, when New Mexico amended a certain form on which the judge was required to indicate that he had informed the defendant, among other things, that a conviction might affect his immigration status, and the defendant and defense counsel, in turn, were required to certify that the defendant had been so advised.