[579]*579LINDER, J.
Petitioner appeals the dismissal of his petition for post-conviction relief, in which he alleged two grounds for relief. First, petitioner claimed that his dangerous offender sentence was imposed unconstitutionally under the principles announced in Apprendi v. New Jersey, 530 US 466, 120 S Ct 2348, 147 L Ed 2d 435 (2000). Second, petitioner claimed that he received constitutionally inadequate assistance of counsel because his trial counsel failed to argue that the dangerous offender sentence was unconstitutional. The trial court dismissed the petition, and we affirm.
The pertinent facts are not in dispute. Petitioner was convicted of first-degree manslaughter after a jury trial in 1989. In August 1989, petitioner was sentenced, pursuant to ORS 161.725 (1987) and ORS 161.735 (1987), as a dangerous offender, based in part on a finding by the sentencing court that he suffers from a severe personality disorder indicating a propensity toward criminal activity. As a result of his dangerous offender status, petitioner received an indeterminate sentence of 30 years, with a 15-year minimum sentence. Had he not been designated a dangerous offender, petitioner’s maximum indeterminate sentence for the crime of first-degree manslaughter would have been 20 years. ORS 163.118(2); ORS 161.605(1). On direct appeal in 1990, this court affirmed petitioner’s conviction without opinion, and the Oregon Supreme Court denied review. State v. Teague, 102 Or App 522, 795 P2d 124, rev den, 310 Or 422 (1990). Following the affirmance of his conviction, petitioner filed a timely petition for post-conviction relief that was denied. In 2000, petitioner filed this second petition.
Petitioner’s principal claim in the current petition is that the imposition of a dangerous offender sentence for his crime violated the Due Process Clause of the Fourteenth Amendment to the United States Constitution, as interpreted and applied in Apprendi, because the predicate facts underlying his dangerous offender sentence enhancement were neither pleaded in the indictment nor found by the jury beyond a reasonable doubt. Petitioner also asserts that he received ineffective assistance of trial counsel in violation of [580]*580state and federal constitutional provisions because his counsel failed to argue that the facts on which the dangerous offender sentence was based should have been pleaded in the indictment and found by the jury beyond a reasonable doubt.
The post-conviction court dismissed the petition on three alternative grounds: (1) the petition was untimely under ORS 138.510(3); (2) petitioner’s allegations were barred by the successive petition bar of ORS 138.550(3); and (3) even if petitioner’s claims were not procedurally barred, the imposition of petitioner’s dangerous offender sentence was not unconstitutional under the legal principles announced in Apprendi. Petitioner asserts on appeal that the post-conviction court erred in each of its conclusions. Defendant superintendent responds that the trial court was correct in each of those conclusions and emphasizes that any one is sufficient to compel affirmance.
By way of supplemental briefing, the parties have addressed the additional issue of whether Apprendi applies retroactively in post-conviction proceedings. Defendant makes two related arguments in that regard. Defendant asserts first that, under the federal Supremacy Clause,1 Oregon courts are obligated to apply federal retroactivity principles in state post-conviction proceedings. Next, relying on the federal retroactivity test articulated in Teague v. Lane, 489 US 288, 109 S Ct 1060, 103 L Ed 2d 334 (1989), defendant urges that the legal principle announced in Apprendi is not retroactive. Petitioner and amicus curiae Oregon Criminal Defense Lawyer’s Association (OCDLA) make several responses to our questions about the retroactivity of the rule of law from Apprendi. Their arguments essentially reduce to two alternative propositions: (1) Teague announced a rule of federal procedure that applies only to federal habeas corpus claims and not to state post-conviction proceedings; (2) in any event, Apprendi satisfies the retroactivity test announced in Teague v. Lane.
As explained at length below, Oregon courts are not obligated to follow federal retroactivity principles in state post-conviction proceedings, but it is well settled that we do [581]*581so for prudential reasons. Consequently, the retroactivity of the rule announced in Apprendi should be tested under Teague v. Lane. Applying that test, we conclude, as has the overwhelming majority of other jurisdictions, that Apprendi does not apply retroactively to collateral review proceedings such as those for post-conviction relief. Because that conclusion completely disposes of petitioner’s challenge to his dangerous offender sentence, we do not decide whether that challenge is untimely under ORS 138.510(3) or is precluded by the successive petition bar of ORS 138.550(3). Finally, we reject petitioner’s ineffective assistance of counsel claim. We address each matter in turn.
I. RETROACTIVITY ANALYSIS IN POST-CONVICTION PROCEEDINGS
The federal Supremacy Clause does not require states to adhere to federal retroactivity principles in determining whether to grant post-conviction relief to Oregon prisoners who rely on newly announced federal constitutional pronouncements. Rather, states are free to apply new federal constitutional pronouncements to a broader range of cases — that is, to give those pronouncements greater retroactive application — than federal law requires of federal courts. That much has long been settled. See, e.g., Johnson v. New Jersey, 384 US 719, 733, 86 S Ct 1772, 16 L Ed 2d 882 (1966); State v. Fair, 263 Or 383, 387-88, 502 P2d 1150 (1972). So, defendant’s position to the contrary is wrong. Rather, as amicus OCDLA correctly argues, it is “up to Oregon to adopt its own rules regarding the effect on collateral review of subsequent changes in the law” and “Oregon has done so[.]” Contrary to amicus OCDLA’s position, however, retroactivity is not determined by ORS 138.510(3) and ORS 138.550(3), which address when a petitioner may raise claims not raised at trial or in a prior post-conviction petition. Those statutes codify principles relating to issue preclusion, not retroactivity. Oregon’s approach to retroactivity is reflected in a nearly 40-year-old body of settled precedent.
To place that precedent — and the retroactivity issue — in perspective, we begin with a bit of background. The United States Constitution requires states to furnish state prisoners with “some clearly defined method by which they [582]*582may raise claims of denial of federal rights.” Young v. Ragen, 337 US 235, 239, 69 S Ct 1073, 93 L Ed 1333 (1948). Before the enactment of Oregon’s current statutory post-conviction scheme, a state prisoner seeking relief for an alleged violation of a federal right was confronted with a confusing array of writs and statutory remedies that often failed to address federal claims adequately. See ORS 138.540(1). See generally Jack G. Collins and Carl R. Neil, The Oregon PostconvictionHearing Act, 39 Or L Rev 337, 337-40 (1960). The Oregon Supreme Court in the 1950s responded to the problem by adopting a more “elastic” approach to the availability of relief through habeas corpus. Huffman v. Alexander, 197 Or 283, 349, 253 P2d 289 (1953). In particular, the court declared that it would “shape our procedural apparatus so as to comply” with the United States Supreme Court’s mandate in Young, id., and would “assimilate its procedure in [habeas] cases, as nearly as may be, to that of the federal courts.” Id. at 350.
4. A few years later, in 1959, the legislature enacted the Post-Conviction Hearing Act. Or Laws 1959, ch 636, §§ 1-24. The Act was intended to eliminate the multiplicity of remedies and to ensure the adequacy of state post-conviction procedures to address federal constitutional defects, as required by federal law. See generally Collins and Neil, 39 Or L Rev at 337-40. To that end, the legislature replaced the common-law writ of habeas corpus with a single uniform statutory proceeding. See Benson v. Gladden, 242 Or 132, 136, 407 P2d 634 (1965), cert den, 384 US 908 (1966). The objective was not to enlarge the grounds for relief but, rather, was to codify grounds previously available and to continue the course set by precedent, while eliminating confusion over which of the several common-law writs was the proper vehicle for relief. See Parker v. Gladden, 245 Or 426, 429, 407 P2d 246 (1965), rev’d on other grounds, 385 US 363, 87 S Ct 468, 17 L Ed 2d 420 (1966); Lerch v. Cupp, 9 Or App 508, 512, 497 P2d 379, rev den (1972); Long v. Cupp, 6 Or App 289, 295, 487 P2d 674, rev den (1971).
Consistently with that general design, the availability of relief under Oregon’s post-conviction statutes depends on two factors. First, a petitioner must seek substantive relief [583]*583that would have been available through a writ of habeas corpus. Bartz v. State of Oregon, 314 Or 353, 361, 839 P2d 217 (1992). That principle is codified in ORS 138.530(2), which effectively provides that the substance of habeas corpus relief remains available to a person who petitions for relief under the post-conviction statutes. Bartz, 314 Or at 361. Second, a petitioner must follow the proper procedures for obtaining that relief. Id. As pertinent here, the statutes codify issue preclusion principles that turn on the availability of prior judicial proceedings to address a petitioner’s claims.2 So, for example, a petitioner ordinarily is procedurally barred from raising a claim that could have been raised at trial and on direct appeal or that was raised and decided. See generally Palmer v. State of Oregon, 318 Or 352, 358, 867 P2d 1368 (1994) (construing ORS 138.550(1)); North v. Cupp, 254 Or 451, 455-57, 461 P2d 271 (1969), cert den, 397 US 1054 (1970) (same). Likewise, a petitioner ordinarily may not raise a claim in an untimely or subsequent petition for post-conviction relief if the claim could have been asserted in a timely initial petition. Bartz, 314 Or at 359. An exception to those procedural bars arises when the claim asserted is one that “could not reasonably have been asserted” in the prior proceeding, which encompasses the pronouncement of a new constitutional principle that could not reasonably have been anticipated and raised at an earlier stage. See generally Long v. Armenakis, 166 Or App 94, 100-02, 999 P2d 461, rev den, 330 Or 361 (2000).
Issue preclusion and retroactivity are distinct concepts and traditionally have been treated as such. From the earliest cases decided under Oregon’s post-conviction statutes, petitioners have been able procedurally to assert claims based on unanticipated and newly announced constitutional principles, but their ability to prevail on those claims has depended on the retroactivity of the constitutional principle [584]*584at work. Such cases are legion. See, e.g., North, 254 Or at 459-62 (one of petitioner’s claims barred by statutory issue preclusion principles; claim based on newly announced Escobedo/Miranda doctrine could be raised but was not a basis for relief because doctrine was not retroactive); Church v. Gladden, 244 Or 308, 311-13, 417 P2d 993 (1966) (similar).3 As we explained in Myers v. Cupp, 49 Or App 691, 621 P2d 579 (1980), rev den, 290 Or 491 (1981), whether a new constitutional rule provides an exception to issue preclusion and also is retroactive are complementary inquiries; a petitioner must satisfy both to be entitled to post-conviction relief:
“[A]n exception to the general rule that an issue raised and considered on direct appeal cannot be reconsidered in a post-conviction proceeding applies where the law with respect to that issue has changed since the time of appeal and that new law is to be applied retroactively.”
49 Or App at 695-96 (emphasis added; citation omitted). In a more recent post-conviction case, the Oregon Supreme Court similarly observed that a “court must analyze retroactivity” [585]*585when considering whether to apply a newly announced rule to a post-conviction case. Moen v. Peterson, 312 Or 503, 508-10, 824 P2d 404 (1991) (emphasis added) (declining to determine what retroactivity standard should apply because constitutional rule invoked by petitioner was not new). In short, assessing the retroactivity of a newly announced constitutional rule is a necessary and legitimate step in the analysis of whether post-conviction relief should be awarded on the basis of that new pronouncement. The inquiry is not subsumed in the statutory standards pertaining to when new issues are procedurally proper to assert.4
Our conclusion that retroactivity principles apply in post-conviction proceedings leads to a second issue: Do we apply federal standards for retroactivity to new federal constitutional pronouncements or do we apply some other standard? That issue also is resolved by precedent.
In Fair, the Oregon Supreme Court reviewed its case law involving the retroactive application of newly announced federal constitutional principles and concluded:
“We may draw two conclusions from our recent decisions on retroactivity. First, we are free to choose the degree of retroactivity or prospectivity which we believe appropriate to the particular rule under consideration, so long as we give federal constitutional rights at least as broad a scope as the United States Supreme Court requires. Secondly, we have tended to restrict the retroactive application of newly - announced rights, giving them only the application which the Supreme Court has adopted as a minimum.”
263 Or at 387-88 (emphasis added). To be sure, Fair was a direct appeal, not a post-conviction case. That fact, however, played no part in the court’s rationale, nor is it reflected as a limitation on its holding. To the contrary, most of the decisions that the court reviewed in Fair to synthesize its approach to retroactivity of newly announced federal rules [586]*586were ones that the court had rendered in post-conviction appeals. See Fair, 263 Or at 387-88.5 Drawing from that jurisprudence, the court identified a general approach to claims based on the retroactive application of newly announced federal rights. See id. Citing Fair, we have acknowledged that we are to adhere to federal precedent in analyzing the retro-activity of newly announced federal constitutional rules asserted as grounds for collateral relief from criminal convictions, including grounds asserted in petitions for post-conviction relief.6
Fair, at a minimum, announces a preferred approach and stands for the proposition that Oregon courts generally should apply federal retroactivity rules to newly announced federal principles, unless we identify a sound reason to depart from them. That approach remains sound and should be followed in this case. As described earlier, the very existence of our statutory post-conviction relief procedures is, in effect, a reflection of principles of comity and federalism: states must respect the federal constitutional rights of their state prisoners. Those same principles were at work in our preexisting habeas jurisprudence, where the Oregon Supreme Court adopted an approach of aligning its habeas corpus procedures, as nearly as possible, with the procedures and standards followed by federal courts. Huffman, 197 Or at 349. The enactment of the post-conviction statutes in 1959 [587]*587was not intended to broaden remedies for state prisoners but to codify and unify them. Consistently with that understanding, and consistently with comity principles, the Oregon Supreme Court announced that it would apply federal retro-activity principles — even as they changed and evolved — to newly announced federal constitutional rights, despite the fact that it did not have to do so as a matter of federal law:
“There is nothing to prevent this court from adopting a more strict rule [of retroactivity] than that enunciated by the United States Supreme Court, but in this field of law it adds nothing to consistent and orderly judicial process.”
State v. Evans, 258 Or 437, 442, 483 P2d 1300 (1971) (overruling prior post-conviction cases that inaccurately predicted the retroactive scope that the United States Supreme Court later gave to Escobedo / Miranda). See also Fair, 263 Or at 387-88 (synthesizing approach from Evans and earlier post-conviction cases). It would be a perversion of the comity principles reflected in state post-conviction procedures, not a service to them, to adopt rules of retroactivity for new federal pronouncements that are broader than those adopted by federal courts, therefore according less respect to the finality of state court judgments than the federal courts themselves require. See generally Teague v. Lane, 489 US at 310 (approach to retroactivity of new federal constitutional rules should respect state interests in the finality of their judgments and should minimize the cost and intrusiveness of forcing states to continually litigate under changing standards the legality of the confinement of state prisoners whose trial and appeals conformed to then-existing constitutional standards). We therefore conclude that the current standards governing retroactivity adopted by the United States Supreme Court appropriately apply to petitioner’s Apprendibased claim in this case.
-II. APPLICATION OF TEAGUE v. LANE
Under Teague v. Lane, newly announced constitutional rules apply retroactively to cases on collateral review only in two narrow circumstances. 489 US at 310-11. The exception relevant here applies to new “watershed rules of criminal procedure that are necessary to the fundamental fairness of the criminal proceeding.” Sawyer v. Smith, 497 US [588]*588227, 241-42, 110 S Ct 2822, 111 L Ed 2d 193 (1990) (internal quotation marks omitted). To meet it, a new rule must not only be “aimed at improving the accuracy of the trial,” it must also “alter our understanding of the bedrock procedural elements essential to the fairness of a proceeding.” Id. at 242 (emphasis in original; internal quotation marks omitted).
The standard is one that few new pronouncements are likely to satisfy. As the Supreme Court itself has observed, new “watershed” rules are inevitably a rarity: “Because we operate from the premise that such procedures would be so central to an accurate determination of innocence or guilt, we believe it unlikely that many such components of basic due process have yet to emerge.” Teague v. Lane, 489 US at 313. So, for example, even so-called structural errors do not necessarily satisfy the exception, because new rules involving such errors in a trial do not necessarily “alter[ ] our understanding of the bedrock” principle involved. Tyler v. Cain, 533 US 656, 666 n 7, 121 S Ct 2478, 150 L Ed 2d 632 (2001) (emphasis in original). In fact, since Teague v. Lane, the Court has not identified a new constitutional rule or proposed new constitutional rule that qualifies under the “watershed” exception. See United States v. Mandanici, 205 F3d 519, 529 (2d Cir 2000), cert den, 531 US 879 (2000), and cert den, _ US __, 122 S Ct 2666 (2002) (summarizing cases).
The Supreme Court’s holding in Apprendi is that a jury, rather than a judge, must determine the facts that support a statutory sentencing enhancement beyond a statutory maximum and that the determination must be by proof beyond a reasonable doubt, rather than a preponderance. The Apprendi rule therefore has two aspects: (1) who must find the facts that enhance the sentence beyond the maximum (a jury), and (2) the standard of proof that applies to the factfinding (beyond a reasonable doubt). Every appellate court in the country that has considered the retroactivity of Apprendi’s holding, save one,7 has concluded that it does not [589]*589meet the “watershed” standard for an exception to the general rule of nonretroactivity. Those cases are correct.
The Supreme Court has identified Gideon u. Wainwright, 372 US 335, 83 S Ct 792, 9 L Ed 2d 799 (1963), which declared for the first time that representation by counsel is fundamental to a fair criminal trial, as the kind of watershed holding that meets the Teague v. Lane exception. Saffle v. Parks, 494 US 484, 495, 110 S Ct 1257, 108 L Ed 2d 415 (1990). Lower courts have observed that Apprendi is not on a par with Gideon and does not otherwise meet the Teague v. Lane exception for several reasons:
• Gideon applies to every felony prosecution, in contrast to Apprendi, which comes into play only in the subset of criminal cases in which a defendant is sentenced beyond the statutory maximum. Apprendi therefore did not announce the kind of sweeping rule contemplated by Teague v. Lane’s exception for watershed rules. United States v. Sanchez-Cervantes, 282 F3d 664, 669 (9th Cir), cert den, _ US _ , 123 S Ct 48 (2002); United States v. Sanders, 247 F3d 139, 150 (4th Cir), cert den, 534 US 1032 (2001).
• The applicable Teague v. Lane exception applies only to those procedures without which the likelihood of an accurate conviction is seriously diminished. Apprendi’s holding does not protect the innocent from erroneous conviction but, rather, protects the guilty from sentences beyond the statutory maximum. Sanchez-Cervantes, 282 F3d at 669-70; United States v. Moss, 252 F3d 993, 999 (8th Cir 2001), cert den, 534 US 1097 (2002).
[590]*590• Apprendi is not a “watershed” rule because judges can constitutionally increase a criminal defendant’s sentence based on facts that a judge finds by a preponderance of the evidence, provided that increase does not exceed the statutory maximum. Sanders, 247 F3d at 150; Moss, 252 F3d at 1000.
• Although Gideon announced a new bedrock principle by holding that the right to counsel is fundamental to any felony criminal trial, Apprendi “merely clarified and extended the scope of a preexisting right — the right to have all convictions supported by proof beyond a reasonable doubt.” United States v. Mora, 293 F3d 1213, 1219 (10th Cir), cert den, _ US _ , 123 S Ct 388 (2002).
Although all of those observations have persuasive force, they are eclipsed by one further factor that the courts have relied on — namely, that Apprendi is subject to harmless error/plain error analysis. See, e.g.,Moss, 252 F3d at 1000-01; Sanders, 247 F3d at 150. Many lower court decisions relied on that factor before the United States Supreme Court addressed the question, effectively predicting that the Court would conclude that Apprendi is subject to harmless error/ plain error analysis. Last term, the Court proved them correct. In United States v. Cotton, 535 US 625, 122 S Ct 1781, 152 L Ed 2d 860 (2002), the Court affirmed a defendant’s enhanced sentence that was based on the quantity of drugs he possessed, notwithstanding an admitted Apprendi error, after concluding that “the error did not seriously affect the fairness, integrity, or public reputation of judicial proceedings.” 122 S Ct at 1786. As other courts have correctly observed, that conclusion is inherently at odds with the conclusion that Apprendi announces a rule that satisfies Teague v. Lane’s exacting standard:
“Further supporting the view that Apprendi does not rise to the level of a watershed change in criminal procedure is the fact that the maj ority of the federal circuit courts have subjected Apprendi claims to harmless and plain error review. * * * As these courts have recognized, it is possible for a criminal defendant to have a fair and accurate trial without the new procedural protection offered by Apprendi. None of these cases have suggested that failure to submit [591]*591the question of [an enhanced sentencing factor] to a jury is structural error. We do not suggest, of course, that all structural errors satisfy Teague’s second exception. We merely emphasize that finding something to be a structural error would seem to be a necessary predicate for a new rule to apply retroactively under Teague.”
Sanders, 247 F3d at 150-51 (citations omitted). There is no persuasive rejoinder to that observation. The Court’s holding in Cotton that Apprendi is subject to a plain error/harmless error analysis effectively dictates a conclusion that Apprendi does not apply retroactively in collateral challenges to a conviction. Curtis v. United States, 294 F3d 841, 843-44 (7th Cir 2002) (so concluding after Supreme Court’s decision in Cotton); Mora, 293 F3d at 1219 (same).8
We therefore hold that Apprendi does not apply retroactively in post-conviction proceedings in this state. As a result, we do not need to decide whether petitioner’s claims were untimely under ORS 138.510(3) or were precluded by the successive petition bar of ORS 138.550(3). Petitioner’s claim that his dangerous offender sentence is unconstitutional under Apprendi was properly dismissed because petitioner may not avail himself of the Apprendi rule in this case.
III. INEFFECTIVE ASSISTANCE OF COUNSEL CLAIM
The only remaining issue is whether the post-conviction court also properly dismissed petitioner’s ineffective assistance of counsel claim. Petitioner’s claim in that regard requires little discussion. To prevail on a claim of ineffective assistance of counsel, petitioner must show that “trial counsel failed to exercise reasonable professional skill and judgment and that [he] suffered prejudice as a result.” Trujillo v. Maass, 312 Or 431, 435, 822 P2d 703 (1991). When a petitioner bases his claim on his attorney’s failure to raise a particular legal position in his criminal trial, the attorney’s representation is not constitutionally inadequate if, due to [592]*592the unsettled state of the law, reasonable counsel could have disagreed as to whether to assert the position. Wells v. Peterson, 315 Or 233, 236, 844 P2d 192 (1992). As the above discussion amply demonstrates, Apprendi plainly announced a new rule of federal constitutional law, even if it was not one of watershed dimensions that applies retroactively to collateral review proceedings under Teague v. Lane. Because Apprendi’s rule was new, counsel exercising reasonable professional skill and judgment could not anticipate that the law would develop as it did, and reasonable counsel could have disagreed as to whether to raise the argument that Apprendi eventually endorsed. For that reason, the post-conviction court correctly dismissed petitioner’s ineffective assistance of counsel claim.
IV. CONCLUSION
In summary, federal retroactivity principles govern petitioner’s ability to assert a claim for post-conviction relief based on the rule of law announced in Apprendi. Under the guidelines set forth in Teague v. Lane, newly announced constitutional rules do not apply retroactively to cases on collateral review unless the newly announced rule is a “watershed rule of criminal procedure.” For the reasons stated above, the rule in Apprendi is not such a rule. Petitioner, therefore, may not avail himself of the Apprendi rule in this post-conviction proceeding. Additionally, the post-conviction court correctly dismissed petitioner’s ineffective assistance of counsel claim.
Affirmed.