OPINION OF THE COURT
RENDELL, Circuit Judge.In Horn v. Banks, 536 U.S. 266, 122 S.Ct. 2147, 158 L.Ed.2d 301 (2002), the United States Supreme Court directed us to analyze whether Mills v. Maryland, 486 U.S. 367, 108 S.Ct. 1860, 100 L.Ed.2d 384 (1988), could be retroactively applied under the principles articulated in Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989), for purposes of our collateral review of George Banks’s conviction and sentence. As a result, the Court reversed that portion of our opinion in Banks v. Horn, 271 F.3d 527 (3d Cir.2001), dealing with Teague. We now conclude that Mills did not announce a new rule of *230constitutional law for retroactivity purposes, and thus that our analysis and resolution of Banks’s Mills claims was proper. Accordingly, we will endorse the reasoning set forth in the remainder of our prior opinion.
I.
George Banks was sentenced to death for the murder of thirteen people in Wilkes-Barre, Pennsylvania, in 1982. His conviction and sentence were upheld by the Supreme Court of Pennsylvania on direct appeal, Commonwealth v. Banks, 513 Pa. 318, 521 A.2d 1 (1987), and on appeal for state post-conviction relief. Commonwealth v. Banks, 540 Pa. 143, 656 A.2d 467 (1995). Banks then sought a writ of habeas corpus in the Middle District of Pennsylvania, which was denied in August of 1999. Banks v. Horn, 63 F.Supp.2d 525 (M.D.Pa.1999).
On October 31, 2001, we reversed the District Court and granted Banks a provisional writ of habeas corpus, finding meritorious Banks’s argument that his death sentence was unconstitutional. Banks v. Horn, 271 F.3d 527 (3d Cir.2001) (“Banks I”). Specifically, we found that the sentencing phase instructions and forms violated Mills v. Maryland, 486 U.S. 367, 108 S.Ct. 1860, 100 L.Ed.2d 384 (1988). In Mills, the United States Supreme Court reversed a death sentence where there was a substantial probability that a reasonable jury could have understood the sentencing instructions and forms to disallow the consideration of mitigating factors not unanimously found to exist. Id. at 384, 108 S.Ct. 1860. In Banks I, we concluded that based on the language of the instructions and verdict slip employed in Banks’s penalty phase, a reasonable possibility existed that the jurors believed they were precluded from considering mitigating evidence they had not found unanimously. Banks I, 271 F.3d at 547-551.
In reaching that conclusion, we were presented with the question of whether Mills was applicable for purposes of our collateral review of Banks’s conviction and sentence under Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989). In Teague, the Supreme Court revolutionized the structure for analyzing the retroactivity of criminal procedure decisions, holding that, with rare exception, prisoners may not rely on “new rules”— essentially, rules not settled by pre-exist-ing precedent — for purposes of federal ha-beas corpus review. Id. at 299-301, 109 S.Ct. 1060. Teague thus directed that new decisions of constitutional criminal procedure that are favorable to a prisoner are usually inapplicable once the prisoner has fully exhausted her direct appeals, including the filing of a writ of certiorari to the United States Supreme Court. Id.
Teague’s new rule of nonretroactivity was premised at least in part on a respect for the workings of state courts and state judges appropriate to our federal system. In particular, the Supreme Court has noted that by validating “reasonable, good-faith interpretations of existing precedents made by state courts,” the principles of nonretroactivity established in Teague “effectuate[ ] the States’ interest in the finality of criminal convictions and fosterf ] comity between federal and state courts.” Gilmore v. Taylor, 508 U.S. 333, 340, 113 S.Ct. 2112, 124 L.Ed.2d 306 (1993); see also Teague, 489 U.S. at 308, 109 S.Ct. 1060.
Because Banks’s conviction became final in October of 1987,1 eight months *231before the Supreme Court issued its decision in Mills, one of the Commonwealth’s primary arguments before us in Banks I was that Mills was not applicable to Banks’s petition for habeas relief. We disagreed. We reasoned that, although Teague “retroactivity is a ‘threshold question,’ ” Banks I, 271 F.3d at 541 (quoting Teague, 489 U.S. at 300, 109 S.Ct. 1060), because the Pennsylvania Supreme Court decision itself applied Mills (albeit doing so unreasonably), neither Teague, nor its underlying purposes, required us to perform a retroactivity analysis of Mills. Banks I, 271 F.3d at 541-43. Instead, we found it necessary only to review the merits of the Pennsylvania Supreme Court’s decision, concluding:
Teague teaches that the federal courts in habeas corpus proceedings should be reluctant to apply new rules of federal jurisprudence in state court cases decided before such new rules were handed down. Principles of comity and finality counsel that we maintain a circumscribed scope of habeas review. Here, however, ... the Pennsylvania Supreme Court applied Mills. We are examining the application of Mills, not because we wish to impose a new rule not considered by the Pennsylvania Supreme Court, but as the court in fact did consider and apply it. In such a situation, Teague is not implicated. Accordingly, we need ask only whether the Pennsylvania Supreme Court’s application of Mills should be disturbed under [the appropriate standard of review].
Id. at 543 (citations omitted). Accordingly, we held that resolution of the retroactivity of Mills under Teague was unnecessary, and proceeded directly to our examination of the merits of the Pennsylvania Supreme Court’s application of Mills to the facts presented in Banks’s appeal. As noted above, we resolved that question in Banks’s favor, holding that the sentencing phase jury instructions and forms were clearly unconstitutional, and therefore that the Pennsylvania Supreme Court’s decision finding otherwise involved an unreasonable application of established Supreme Court precedent. Id. at 551.
In Horn v. Banks, 536 U.S. 266, 122 S.Ct. 2147, 2148, 153 L.Ed.2d 301 (2002) (“Banks II”), the Supreme Court concluded otherwise, explicitly and emphatically holding that “federal courts must address the Teague question when it is properly argued by the government.” In doing so, the Court focused on its statements in Caspari v. Bohlen, 510 U.S. 383, 114 S.Ct. 948, 127 L.Ed.2d 236 (1994), that Teague’s “nonretroactivity principle prevents a federal court from granting habeas corpus relief to a state prisoner based on a” new rule, and thus that “if the State ... argue[s] that the defendant seeks the benefit of a new rule of constitutional law, the court must apply Teague before considering the merits of the claim.” Id. at 389, 114 S.Ct. 948 (emphasis in original). Applying these principles, the Supreme Court found that it was “incumbent upon” us to “perform a Teague analysis before granting respondent relief under Mills,” and that we “erred in concluding that [we] did ‘not need to focus on anything other than the reasoning and determination of the Pennsylvania Supreme Court.’ ” Banks II, 122 S.Ct. at 2150 (quoting Banks I, 271 F.3d at 541). Accordingly, the Supreme Court “reverse[d][our] holding that ‘Teag-ue is not implicated’ by this case, and remand[ed] for further proceedings consis*232tent with” its decision. Banks II, 122 S.Ct. at 2151 (quoting Banks I, 271 F.3d at 543).2
II.
We note at the outset that our determination as to the merits of Banks’s Mills claim was not reviewed by the Supreme Court. The Court thus did not vacate our previous decision but only reversed that portion of our opinion that concluded that a Teague analysis was unnecessary for our review of Banks’s habeas petition. Accordingly, the sole issue presently before us is whether our application of Mills on habeas review of Banks’s sentence was improper under the Supreme Court’s non-retroactivity jurisprudence.3 To provide background for the analysis, we first briefly discuss the Court’s decision in Mills itself, then turn to an examination of the Supreme Court’s retroactivity framework.
A.
In Mills, the Court considered the constitutionality of a set of jury instructions, as well as the implementing verdict forms, that could be understood to prevent the consideration of mitigating circumstances if the jury was not unanimous in finding the existence of such circumstances. Mills, 486 U.S. at 371, 108 S.Ct. 1860. That is, “even if some or all of the jurors were to believe some mitigating circumstance or circumstances were present, unless they could unanimously agree on the existence of the same mitigating factor, the sentence necessarily would be death.”4 Id. (emphasis in original). The Court cited the following two possibilities as constitutionally problematic:
If eleven jurors agree that there are six mitigating circumstances, the result is that no mitigating circumstance is found. Consequently, there is nothing to weigh against any aggravating circumstance found and the judgment is death even though eleven jurors think the death penalty wholly inappropriate....
[In] a situation just as intuitively disturbing: All 12 jurors might agree that some mitigating circumstances were present, and even that those mitigating circumstances were significant enough to outweigh any aggravating circumstance found to exist. But unless all 12 could agree that the same mitigating circumstance was present, they would never be permitted to engage in the weighing process or any deliberation on the appropriateness of the death penalty-
Id. at 373-74, 108 S.Ct. 1860 (citations omitted). Noting that imposition of the death penalty under such circumstances *233would “be the height of arbitrariness,” id. at 374, 108 S.Ct. 1860, it went on to state:
It is beyond dispute that in a capital case the sentencer may not be precluded from considering, as a mitigating factor, any aspect of a defendant’s character or record and any of the circumstances of the offense that the defendant proffers as a basis for a sentence less than death. The corollary that the sentencer may not refuse to consider or be precluded from considering any relevant mitigating evidence is equally well established.
Id. at 374-75, 108 S.Ct. 1860 (first and third emphasis added) (quotations and citations omitted). The Court then reiterated the constitutional problem at hand: “[I]f petitioner is correct, a jury that does not unanimously agree on the existence of any mitigating circumstance may not give mitigating evidence any effect whatsoever, and must impose the sentence of death.” Id. at 375, 108 S.Ct. 1860. The Court stated that its existing jurisprudence prohibited any “barrier to the sentencer’s consideration of all mitigating evidence .... [wjhatever [its] cause.” Id. As the Court found a “substantial probability” that reasonable jurors would have understood themselves as being precluded from considering mitigating evidence not found unanimously, the Court reversed Mills’s sentence of death, concluding: “Under our cases, the sentencer must be permitted to consider all mitigating evidence. The possibility that a single juror could block such consideration, and consequently require the jury to impose the death penalty, is one we dare not risk.” Id. at 384, 108 S.Ct. 1860.
B.
Again, the sole issue before us is whether the rule enunciated in Mills is retroactively applicable to Banks’s appeal. Retroactivity analysis is governed by the principles first articulated in Teague v. Lane, in which the Supreme Court held that “[u]nless they fall within an exception to the general rule, new constitutional rules of criminal procedure will not be applicable to those cases which have become final before the new rules are announced.” Teague, 489 U.S. at 310, 109 S.Ct. 1060. Application of this principle of retroactivity proceeds in three steps. See, e.g., Caspari, 510 U.S. at 390, 114 S.Ct. 948. First, we must determine when the defendant’s conviction became final. Id. Second, we must survey the legal landscape to determine whether or not the case in question announced a new rule of constitutional law. Id. Finally, if we determine that the case did announce a new rule, we must consider whether it fits into one of the two exceptions to nonretroactivity. Id. Those exceptions are reserved for (1) rules that “placet ] a class of private conduct beyond the power of the State to proscribe, ... or address[ ] a substantive categorical guarantee accorded by the Constitution, such as a rule prohibiting a certain category of punishment for a class of defendants because of their status or offense,” or (2) “watershed rules of criminal procedure impheating the fundamental fairness and accuracy of the criminal proceeding.” Saffle v. Parks, 494 U.S. 484, 494-95, 110 S.Ct. 1257, 108 L.Ed.2d 415 (1990) (citations and internal quotations omitted). Because the exceptions find rather narrow applicability, however, the typically dispositive step in the Teague retroactivity analysis is the determination of whether the implicated constitutional principle qualifies as a “new rule.”
In Teague itself, the Court admitted that it is “often difficult to determine” whether a case announces a new rule, and explicitly avoided any “attempt to define the spectrum of what may or may not constitute a new rule for retroactivity purposes.” Teague, 489 U.S. at 301, 109 S.Ct. 1060. *234The Supreme Court has further recognized that the inquiry is particularly difficult where the decision in question merely extended the reasoning of prior cases. See, e.g., Saffle, 494 U.S. at 488, 110 S.Ct. 1257; Graham v. Collins, 506 U.S. 461, 467, 113 S.Ct. 892, 122 L.Ed.2d 260 (1993); Butler v. McKellar, 494 U.S. 407, 412-13, 110 S.Ct. 1212, 108 L.Ed.2d 347 (1990); see also Mackey v. United States, 401 U.S. 667, 695, 91 S.Ct. 1160, 28 L.Ed.2d 404 (1971) (Harlan, J., concurring in judgments in part and dissenting in part) (noting the “inevitable difficulties” in distinguishing new rules from the application of old rules in analogous cases).
Perhaps as a result of the inevitable difficulty in articulating one test to govern all possible retroactivity scenarios, the Supreme Court has “stated variously the formula for determining when a rule is new.” O’Dell v. Netherlands, 521 U.S. 151, 156, 117 S.Ct. 1969, 138 L.Ed.2d 351 (1997). Under the Court’s original explication in Teague, “a case announces a new rule when it breaks new ground or imposes a new obligation on the States or the Federal Government.” Teague, 489 U.S. at 301, 109 S.Ct. 1060 (emphasis added). “To put it differently,” the Teague Court explained, “a case announces a new rule if the result was not dictated by precedent existing at the time the defendant’s conviction became final.” Id. (emphasis in original); see also, e.g., Graham, 506 U.S. at 467, 113 S.Ct. 892 (“A holding constitutes a new rule within the meaning of Teague if it breaks new ground, imposes a new obligation on the States or the Federal Government, or was not dictated by precedent existing at the time the defendant’s conviction became final.” (quotations omitted) (emphasis in original)). Similarly, the Court has stated that previous precedents must not simply “inform, or even control or govern” the analysis, but instead must “compel the rule” sought by the defendant. Saffle, 494 U.S. at 491, 110 S.Ct. 1257; see also Butler, 494 U.S. at 415, 110 S.Ct. 1212 (noting that it is insufficient that a decision was considered to be controlled or governed by prior opinions).
At the same time, the Court has focused on the decision-making process confronting state court judges. See, e.g., O’Dell, 521 U.S. at 156, 117 S.Ct. 1969 (“At bottom, ... the Teague doctrine ‘validates reasonable, good-faith interpretations of existing precedents made by state courts even though they are shown to be contrary to later decisions.’ ” (quoting Butler, 494 U.S. at 414, 110 S.Ct. 1212)); Graham, 506 U.S. at 467, 113 S.Ct. 892; Teague, 489 U.S. at 308, 109 S.Ct. 1060. And, in recent decisions, the Court has approached the inquiry from the standpoint of a “reasonable jurist.” In Lambrix v. Singletary, 520 U.S. 518, 117 S.Ct. 1517, 137 L.Ed.2d 771 (1997), for instance, the Court asked whether the “unlawfulness of [the defendant’s] conviction was apparent to all reasonable jurists.” Id. at 527-28, 117 S.Ct. 1517; see also id. at 531, 117 S.Ct. 1517 (examining whether a “reasonable jurist ... could have reached a conclusion different from” the one ultimately reached by the Supreme Court); id. at 526, 117 S.Ct. 1517 (stating that our inquiry is to “determine whether a state court considering the defendant’s claim at the time his conviction became final would have felt compelled by existing precedent to conclude that the rule he seeks was required by the Constitution” (quotations omitted)); O’Dell, 521 U.S. at 166, 117 S.Ct. 1969 (“Teague asks state court judges to judge reasonably, not presciently.”). In sum, “unless reasonable jurists hearing petitioner’s claim at the time his conviction became final ‘would have felt compelled by existing precedent’ to rule in his favor, we are barred from doing so now.” Graham, 506 U.S. at 467, *235113 S.Ct. 892 (quoting Saffle, 494 U.S. at 488, 110 S.Ct. 1257).
III.
We hold that Mills did not announce a new rule of constitutional law for retroactivity purposes, and accordingly that our application of Mills on our habeas review of Banks’s sentence was completely proper.5 There are four aspects to our reasoning: (1) the legal landscape at the time of Banks’s conviction, (2) the Supreme Court’s decision in Mills itself, (3) the relevant post-Mills decisions of the Supreme Court, and (4) the opinions of our sister Courts of Appeals who have addressed whether Teague bars retroactive application of Mills.
Our “first and principal task” under Teague is to survey the legal landscape to determine whether Mills “was dictated by then existing precedent ... that is, [whether] the unlawfulness [of the situation in Mills] was apparent to all reasonable jurists.” Lambrix, 520 U.S. at 527-28, 117 S.Ct. 1517. As discussed above, in Mills the Supreme Court reversed a sentence of death where there was “a substantial probability that reasonable jurors, upon receiving the judge’s instructions in th[e] case, and in attempting to complete the verdict form as instructed, ... thought they were precluded from considering any mitigating evidence unless all 12 jurors agreed on the existence of’ any particular circumstance. Mills, 486 U.S. at 384, 108 S.Ct. 1860. We find highly persuasive Banks’s argument that, given the legal landscape, Mills represented merely an application of the well established constitutional rule that the Eighth Amendment prohibits all barriers to the sentencer’s consideration of any and all mitigation evidence in the penalty phase of a capital trial.
By the time Banks’s conviction became final in 1987, the legal landscape was primarily defined by Supreme Court case law spanning nearly a dozen years.6 We begin our examination of this precedent with the Supreme Court’s decision in Woodson v. North Carolina, 428 U.S. 280, 96 S.Ct. 2978, 49 L.Ed.2d 944 (1976), which struck down North Carolina’s mandatory death penalty statute. Of the many constitutional flaws the plurality found in North Carolina’s capital sentencing structure,7 one *236particularly notable defect was its “failure to allow the particularized consideration of relevant aspects of the character and record of each convicted defendant before the imposition upon him of a sentence of death.” Woodson, 428 U.S. at 303, 96 S.Ct. 2978; see also Roberts (Stanislaus) v. Louisiana, 428 U.S. 325, 96 S.Ct. 3001, 49 L.Ed.2d 974 (1977) (striking down Louisiana’s mandatory death penalty statute). The plurality reiterated that death as a penalty is distinguishable in kind from all other penalties, and held that “the fundamental respect for humanity underlying the Eighth Amendment requires consideration of the character and record of the individual offender and the circumstances of the particular offense as a constitutionally indispensable part of the process of inflicting the penalty of death.” Woodson, 428 U.S. at 304, 96 S.Ct. 2978 (citations omitted); see also Roberts, 428 U.S. at 333-34, 96 S.Ct. 3001 (plurality opinion) (noting that the Constitution requires a “focus on the circumstances of the particular offense and the character and propensities of the offender”); Jurek v. Texas, 428 U.S. 262, 271, 96 S.Ct. 2950, 49 L.Ed.2d 929 (1976) (plurality opinion) (stating that the Eighth and Fourteenth Amendments require that the sentencer be allowed to consider mitigating circumstances).
The Court articulated the full import of Woodson’s constitutional directive more clearly in Lockett v. Ohio, 438 U.S. 586, 98 S.Ct. 2954, 57 L.Ed.2d 973 (1978), in which it struck down a statute that restricted the range of mitigating factors that could be considered by a jury.8 Whereas Woodson involved a statute precluding any consideration of mitigating evidence, Lockett involved somewhat more complex questions: “which facets of an offender or his offense [are] relevant in capital sentencing,” and “what degree of consideration of relevant facets” does the Constitution require. Id. at 604, 98 S.Ct. 2954. The Court responded, in expansive language, that:
[T]he Eighth and Fourteenth Amendments require that the sentencer ... not be precluded from considering, as a mitigating factor, any aspect of a defendant’s character or record and any of the circumstances of the offense that the defendant proffers as a basis for a sentence less than death.
Id. (fust emphasis added). The Court explained that:
Given that the imposition of death by public authority is so profoundly different from all other penalties, we cannot avoid the conclusion that an individualized decision is essential in capital cases. The need for treating each defendant in a capital case with that degree of respect due the uniqueness of the individual is far more important than in noncapi-tal cases .
There is no perfect procedure for deciding in which cases governmental authority should be used to impose death. But a statute that prevents the sentencer in all capital cases from giving independent mitigating weight to aspects of the defendant’s character and record and to circumstances of the offense proffered in mitigation creates the risk that the death penalty will be imposed in spite of factors which may call for a less severe penalty. When the choice is between life and death, that risk is unacceptable and *237incompatible with the commands of the Eighth and Fourteenth Amendments.
Id. at 605, 98 S.Ct. 2954 (emphasis added). In Lockett, the Court stated in unequivocal terms that to “meet constitutional requirements, a death penalty statute must not preclude consideration of relevant mitigating factors.” Id. at 608, 98 S.Ct. 2954.
Four years later, in Eddings v. Oklahoma, 455 U.S. 104, 102 S.Ct. 869, 71 L.Ed.2d 1 (1982), the constitutional rule articulated in Lockett was first adopted and applied by a majority of the Court. In Eddings, the statute in question allowed defendants to present evidence of any mitigating circumstance — unlike the statute at issue in Lockett — but the trial judge found that he was unable to consider certain mitigating evidence as a matter of law. Id. at 113, 102 S.Ct. 869. The Supreme Court reversed, describing Lockett as requiring that “the sentencer in capital cases must be permitted to consider any relevant mitigating factor,” id. at 112, 102 S.Ct. 869, and holding that:
The limitations placed by these courts upon the mitigating evidence they would consider violated the rule in Lockett. Just as the State may not by statute preclude the sentencer from considering any mitigating factor, neither may the sentencer refuse to consider, as a matter of law, any relevant mitigating evidence. In this instance, it was as if the trial judge had instructed the jury to disregard the mitigating evidence proffered on his behalf. The sentencer ... may determine the weight to be given relevant mitigating evidence. But they may not give it no weight by excluding such evidence from their consideration.
Id. at 113-115, 102 S.Ct. 869 (emphasis added). Thus, in Eddings, the Lockett plurality’s constitutional rule was solidified as a settled and prominent feature of the Supreme Court’s death penalty jurisprudence; indeed, the Court thereafter characterized the rule as one of the two prerequisites to a valid death sentence imposed by the Eighth Amendment. See California v. Brown, 479 U.S. 538, 541, 107 S.Ct. 837, 93 L.Ed.2d 934 (1987); see also Penry v. Lynaugh, 492 U.S. 302, 318, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989) (stating that by early 1986 “it was dear from Lockett and Eddings” that the Constitution prohibited a State from “prevent[ing] the sentencer from considering and giving effect to evidence relevant to the defendant’s background or character or to the circumstances of the offense that mitigate against imposing the death penalty”). The legal landscape at the time of Banks’s conviction, however, was further shaped by three additional cases in which the Court had occasion to apply the Lockett/Eddings rule.
First, in Skipper v. South Carolina, 476 U.S. 1, 106 S.Ct. 1669, 90 L.Ed.2d 1 (1986), the Supreme Court relied on Lockett and Eddings to reverse the defendant’s death sentence after the trial judge ruled that certain mitigating evidence was inadmissible and prohibited the sentencing jury from considering it. The Court began by reiterating its “well established” Lockett/Eddings rule, stating:
There is no disputing that ... in capital cases the sentencer may not be precluded from considering, as a mitigating factor, any aspect of a defendant’s character or record and any of the circumstances of the offense that the defendant proffers as a basis for a sentence less than death. Equally clear is the corollary rule that the sentencer may not refuse to consider or be precluded from considering any mitigating evidence.
Id. at 4, 106 S.Ct. 1669 (quotations and citations omitted). The Court then addressed the sole question before it: *238“whether the exclusion from the sentencing hearing of the testimony petitioner proffered ... deprived petitioner of his right to place before the sentencer relevant evidence in mitigation of punishment.” Id. Noting that it could “hardly be disputed” that the exclusion did have that effect, id., the Court concluded that “[t]he exclusion by the state trial court of relevant mitigating evidence impeded the sentencing jury’s ability to carry out its task of considering all relevant facets of the character and record of the individual offender.” Id. at 8, 106 S.Ct. 1669.
A year later, in California v. Brown, 479 U.S. 538, 107 S.Ct. 837, 93 L.Ed.2d 934 (1987), the Court again affirmed the constitutional principles established in Lockett and Eddings, but this time upheld the underlying death sentence.9 In Brown, the defendant challenged the constitutionality of “an instruction informing jurors that they ‘must not be swayed by mere sentiment, conjecture, sympathy, passion, prejudice, public opinion or public feeling.’” Brown, 479 U.S. at 540, 107 S.Ct. 837. The Court began by reiterating that the Eighth Amendment requires that capital defendants be allowed to introduce any relevant mitigating evidence, and further that consideration of such evidence is a “ ‘constitutionally indispensable part of the process of inflicting the penalty of death.’ ” Brown, 479 U.S. at 541, 107 S.Ct. 837 (quoting Woodson, 428 U.S. at 304, 96 S.Ct. 2978). Applying these principles to the instruction before them, however, the Court found that it merely “prohibit[ed] juries from basing their sentencing decisions on factors not presented at the trial, and irrelevant to the issues at the trial,” limitations fully consistent with the Constitution’s requirement that the jury be allowed to consider any mitigating evidence. Brown, 479 U.S. at 543, 107 S.Ct. 837; see also Saffle, 494 U.S. at 488-95, 110 S.Ct. 1257 (holding that Teague precluded the defendant’s constitutional challenge to a jury instruction requiring jurors to avoid the influence of sympathy in sentencing).
Finally, in Hitchcock v. Dugger, 481 U.S. 393, 107 S.Ct. 1821, 95 L.Ed.2d 347 (1987), the Supreme Court unanimously reversed the defendant’s death sentence where an “advisory jury was instructed not to consider, and the sentencing judge refused to consider, evidence of nonstatutory mitigating circumstances.” Id. at 398-99, 107 S.Ct. 1821. The Court held that such a circumstance “did not comport with the requirements of Skipper, Eddings, and Lockett," id. at 399, 107 S.Ct. 1821 (citations omitted), which established that “the sentencer may not refuse to consider or be precluded from considering any relevant mitigating evidence.” Id. at 394, 107 S.Ct. 1821 (quotations omitted). The Court concluded, “[0]ur cases hold that the exclusion of mitigating evidence of the sort at issue here renders the death sentence invalid.” Id. at 399, 107 S.Ct. 1821; see also Burger v. Kemp, 483 U.S. 776, 790 n. 7, 107 S.Ct. 3114, 97 L.Ed.2d 638 (1987) (affirming, in dicta, the constitutional principles established in Lockett and Eddings); Sumner v. Shuman, 483 U.S. 66, 75-76, 107 S.Ct. 2716, 97 L.Ed.2d 56 (1987) (same).
Although Banks relies primarily on the Lockett/Eddings line of cases, we mention another Supreme Court case, Andres v. United States, 333 U.S. 740, 68 S.Ct. 880, 92 L.Ed. 1055 (1948), that was cited in Mills and relied on by another Court of Appeals in deciding this issue. See Mills, 486 U.S. at 377 & n. 9, 108 S.Ct. 1860; *239Gall v. Parker, 231 F.3d 265, 323 (6th Cir.2001). Andres involved an assessment and interpretation of a federal death penalty statute, and the question of whether a trial court’s unanimity instructions pursuant to that statute were erroneous. Id. at 746, 68 S.Ct. 880. The statutory scheme was structured such that the finding of guilt meant the automatic imposition of the death penalty unless the verdict was qualified by the phrase “without capital punishment.” Id. The government argued that the jury’s determination as to guilt was conclusive as to the death penalty unless the jury then unanimously decided to qualify the verdict. Id. The Supreme Court opined, however, that the proper construction required that the “jury’s decision upon both guilt and whether the punishment of death should be imposed must be unanimous.” Id. at 749, 68 S.Ct. 880. That is, the Court required that the jury consider and be unanimous that death should be the penalty imposed. Although it recognized that “the interpretation ... urged by the Government cannot be proven erroneous with certainty,” the Court found its construction “more consonant with the general humanitarian purpose of the statute and the history of the Anglo-American jury system.” Id. at 748-49, 68 S.Ct. 880. The Court then held that the instruction given to the jury by the District Court conveyed the erroneous interpretation of the statute. Id. at 749-52, 68 S.Ct. 880. Under the instructions given, the Court concluded, the jury might have erroneously but “reasonably conclude[d] that, if they [could not] all agree to grant mercy, the verdict of guilt must stand unqualified.” Id. at 752, 68 S.Ct. 880. Accordingly, the Court overturned the sentence and ordered a new trial. Id.
Clearly, the Andres decision took place within a different statutory context and, in our view, is not a model of clarity. However, because it, like Mills, involved jury instructions on unanimity in a capital case, Andres certainly plays some role in the relevant legal landscape — a conclusion reinforced by the Court’s citation to it in Mills itself. See Mills, 486 U.S. at 377 & n. 9, 108 S.Ct. 1860. At the very least, Andres invokes a number of themes that, significantly, are consistent with and complimentary to the Court’s later constitutional death penalty jurisprudence. For instance, Andres indicates a concern for particular clarity in capital jury instructions, noting that doubts “should be resolved in favor of the accused.” Andres, 333 U.S. at 752, 68 S.Ct. 880; see also Mills, 486 U.S. at 377, 108 S.Ct. 1860. Further, it is particularly noteworthy that the upshot of the Court’s decision was that the statute and jury instructions were interpreted to avoid a situation in which a juror could be prevented — by operation of a requirement for unanimity to avoid the death penalty, and, accordingly, by the views of other jurors — from giving effect to his or her belief that death was an inappropriate sentence under particular circumstances. Andres, 333 U.S. at 748-52, 68 S.Ct. 880.
We agree with Banks that this legal landscape — as exemplified by Lockett and Eddings, but also including at least Andres, Woodson, Skipper, Brown, and Hitchcock — strongly supports Banks’s position that the Court in Mills did not develop any new principle of law, but instead merely relied upon clear and well established constitutional rules, such that Mills was compelled and dictated by the legal landscape, and no reasonable jurist could have reached a different result. Insofar as the landscape evidenced the Supreme Court’s unwavering recognition and insistence that the Eighth Amendment prohibits any barrier to the sentencer’s consideration of mitigating evidence, it provided a clear indication that a jury instruction that *240could work to prevent a juror from considering any and all mitigating evidence, whether because of unanimity requirements or otherwise, would be constitutionally infirm.
In reaching this conclusion, we are not unmindful of the difficulty of employing Teague’s mandates to divine whether the legal landscape supports a finding that a rule is or is not new, given the Court’s various formulations of the measuring stick for determining whether a particular case does or does not announce a new rule. It is not precisely clear just how short the “step” must be between existing precedent and the current announcement, or how strong the pull of precedent must be in a certain direction. There certainly must be some gradation or difference, or the rule in question would not be even arguably new. Thus, a decision that does extend reasoning may nonetheless be viewed as not “new” under Teague. The Supreme Court has acknowledged as much when it has noted the difficulty of determining whether a new rule was announced where “a decision extends the reasoning of our prior cases.” Saffle, 494 U.S. at 488, 110 S.Ct. 1257; see also Graham, 506 U.S. at 467, 113 S.Ct. 892; Butler, 494 U.S. at 412-13, 110 S.Ct. 1212. To read certain of the operative terms the Court has employed, such as “dictated” and “commanded,” narrowly, such that they would require express direction from the existing precedent, would be to unrealistically require courts to have anticipated all future scenarios in order for later cases to not announce a new rule. Another term the Court has used, “compel,” has been defined as to “force, drive, [or] impel.” Webster’s Third New Int’l Dictionary 463 (1993). This seems not only to be a more functional description of the test, but it also fits nicely with the concept of the “reasonable jurist” that is referenced in many of the Court’s recent cases in this area. That is, we ask whether the existing precedent set forth a rule that all reasonable jurists would agree impels or drives the result in the new situation presented. Here, the existing case law clearly provided that sentencers could not be prevented from considering any and all mitigating evidence. In Mills, the Court merely recognized that the perceived need for unanimity could constitute one such unconstitutional barrier. Even if one were to question whether the result was “dictated” or “commanded” by the constitutional rule itself, it surely was compelled in the sense that previous pronouncements would constrain all reasonable jurists to conclude the situation in Mills to be unconstitutional. It is perhaps this shading that distinguishes our view of Mills from that of our concurring colleague.10
The Supreme Court’s reasoning and rhetoric in Mills itself follows form from the legal landscape, and bolsters the view that it was not announcing a new constitutional rule. Initially, we note that “[i]t is significant” that Mills did explicitly and heavily rely on controlling precedent. Lambrix, 520 U.S. at 528, 117 S.Ct. 1517. Unlike the situation in Lambrix, in which the Supreme Court found a “new rule” in part because the underlying decision cited only a single case — and with a “cf.” signal at that, see id. at 528-29, 117 S.Ct. 1517 — Mills is replete with references to controlling precedent; the Court frequently cited to and quoted from Lockett, Eddings, Skipper, and Hitchcock. See Mills, 486 U.S. at 374-76, 108 S.Ct. 1860. Moreover, the Court used language in its analysis that does not merely state, but indeed exhorts, the rich precedent compelling its *241reasoning and result. The Court’s precise wording bears repeating here:
It would certainly be the height of arbitrariness to allow or require the imposition of the death penalty under the circumstances ... postulated by petitioner .... It is beyond dispute that in a capital case “ ‘the sentencer [may] not be precluded from considering, as a mitigating factor, any aspect of a defendant’s character or record and any of the circumstances of the offense that the defendant proffers as a basis for a sentence less than death.’ ” Eddings v. Oklahoma, 455 U.S. 104, 110, 102 S.Ct. 869, 874, 71 L.Ed.2d 1 (1982), quoting Lockett v. Ohio, 438 U.S. 586, 604, 98 S.Ct. 2954, 2964-2965, 57 L.Ed.2d 973 (1978) (plurality opinion) (emphasis in original). See Skipper v. South Carolina, 476 U.S. 1, 4, 106 S.Ct. 1669, 1670-1671, 90 L.Ed.2d 1 (1986). The corollary that “the sentencer may not refuse to consider or be precluded from considering ‘any relevant mitigating evidence’ ” is equally “well established.” Ibid. (emphasis added), quoting Eddings, 455 U.S., at 114, 102 S.Ct., at 877.
Under our decisions, it is not relevant whether the barrier to the sentencer’s consideration of all mitigating evidence is interposed by statute, Lockett v. Ohio, supra; Hitchcock v. Dugger, 481 U.S. 393, 107 S.Ct. 1821, 95 L.Ed.2d 347 (1987); by the sentencing court, Eddings v. Oklahoma, supra; or by an evidentiary ruling, Skipper v. South Carolina, supra. The same must be true with respect to a single juror’s holdout vote against finding the presence of a mitigating circumstance: Whatever the cause, if petitioner’s interpretation of the sentencing process is correct, the conclusion would necessarily be the same: “Because the [sentencer’s] failure to consider all of the mitigating evidence risks erroneous imposition of the death sentence, in plain violation of Lockett, it is our duty to remand this case for resentencing.” Eddings v. Oklahoma, 455 U.S., at 117, n., 102 S.Ct., at 878, n. (O’Connor, J., concurring).
The critical question, then, is whether petitioner’s interpretation of the sentencing process is one a reasonable jury could have drawn from the instructions given by the trial judge and from the verdict form employed in this case.
Mills, 486 U.S. at 374-76, 108 S.Ct. 1860. Given this language, it can hardly be disputed that Mills did not announce a new rule.11 As the Court recognized, it was well established by the Lockett/Eddings line of cases that the Constitution prohibited any barrier to the jury’s consideration of mitigating evidence, “[w]hatever the cause.” Id. at 375, 108 S.Ct. 1860. The relevant cases had never indicated that the source or form some particular barrier took would be relevant to its constitutionality; instead they had consistently and repeatedly prohibited, in clear language, any barrier. See, e.g., Hitchcock, 481 U.S. at 394, 107 S.Ct. 1821 (“[I]n capital cases, the sentencer may not refuse to consider or be precluded from considering any relevant mitigating evidence.” (quotations omitted)); Skipper, 476 U.S. at 4, 106 S.Ct. 1669 (same); see also, e.g., Saffle, 494 U.S. at 491, 110 S.Ct. 1257 (“Lockett and Ed*242dings command that the State must allow the jury to give effect to mitigating evidence in making the sentencing decision.”). The extent of the relevant holding in Mills, then, was merely its acknowledgment of a conclusion already required by the governing constitutional rules: that if a jury instruction and verdict form, because of its unanimity requirements, precluded juror consideration of any and all mitigation evidence, the resulting death sentence would be unconstitutional. See McKoy v. North Carolina, 494 U.S. 433, 438, 110 S.Ct. 1227, 108 L.Ed.2d 369 (1990) (“[Ajllowing a ‘holdout’ juror to prevent the other jurors from considering mitigating evidence violated the principle established in Lockett v. Ohio, that a sentencer may not be precluded from giving effect to all mitigating evidence.” (citation omitted)). That these constitutional principles were settled before Mills is further evidenced by the Maryland Court of Appeals decision reversed by the Supreme Court in Mills itself, as the disagreement between the Court of Appeals’ majority and dissenting opinions was unrelated to the underlying constitutional rules that govern capital sentencing. See Mills, 527 A.2d at 13; id. at 33, 527 A.2d 3 (McAuliffe, J., dissenting) (“[T]he majority and I are in essential agreement as to the basic principles of law that control in a capital sentencing proceeding.... A mitigating circumstance ... must be considered by each juror who believes it to have been proven to exist, irrespective of whether all jurors agree that it exists.”); Mills, 486 U.S. at 372, 108 S.Ct. 1860 (“The [Maryland Court of Appeals] did not dispute that if the statute and form were read as petitioner suggested, jurors would be improperly prevented from giving due consideration to mitigating evidence.” (emphasis in original)); see also Butler, 494 U.S. at 415, 110 S.Ct. 1212 (indicating that actual disagreement on a constitutional rule provides some evidence that it is “new” under Teague).
Moreover, we note that recent Supreme Court references to the Teague test have indicated that the “determinative question” under Teague is whether reasonable jurists, reading the case law in existence at the time the conviction became final, could have concluded that Banks’s sentencing “was not constitutionally infirm.” Graham, 506 U.S. at 477, 113 S.Ct. 892 (emphasis in original); see also id. at 476, 113 S.Ct. 892 (“The result in a given case is not dictated by precedent if it is susceptible to debate among reasonable minds, or, put differently, if reasonable jurists may disagree.”) (quoting Stringer v. Black, 503 U.S. 222, 238, 112 S.Ct. 1130, 117 L.Ed.2d 367 (1992) (Souter, J., dissenting)). Given the Supreme Court’s reliance on the Lockett/Eddings rule as an established requirement of the Eighth Amendment by the time Banks’s conviction became final, we are persuaded that the relevant rule in Mills was “apparent,” and that no “reasonable jurist” could have reached a different conclusion. Lambrix, 520 U.S. at 528, 117 S.Ct. 1517; see also, e.g., Butler, 494 U.S. at 415, 110 S.Ct. 1212 (asking whether the relevant outcome was “susceptible to debate among reasonable minds”). Indeed, as the settled Lockett/Eddings rule was a blanket prohibition on barriers to the jury’s consideration of mitigating circumstances, we find ourselves unable to construct any analytic framework, consistent with the legal landscape, under which Mills could have come out differently. Cf. Lambrix, 520 U.S. at 532, 117 S.Ct. 1517 (“There were at least three different ... approaches that would have suggested a different outcome.”). A failure to decide Mills as the Court in fact decided it would not just have taken an “illogical” or “grudging” application of the Lockett/Eddings rule, Butler, 494 U.S. at 415, 110 S.Ct. 1212, it would have taken a complete*243ly untenable one. Any reasonable jurist “would have felt compelled” to decide Mills accordingly. O’Dell, 521 U.S. at 156, 117 S.Ct. 1969 (quoting Lambrix, 520 U.S. at 527, 117 S.Ct. 1517).
While the Commonwealth and our concurring colleague seize on the fact that there were four dissenting justices in Mills in arguing that “reasonable jurists” could have differed on the outcome at the time of that case, a careful reading of the Mills dissent makes clear that the dissenting justices did not take issue with the principle that jurors must be able to consider all mitigating factors without the requirement of unanimity. The dissenters never questioned the strong statements by the majority to the effect that the result in Mills was mandated by the Lockett/Eddings line of cases. Rather, they viewed the majority as having applied the wrong test as to the 'probability that jurors misunderstood the instructions in the factual setting presented. See Mills, 486 U.S. at 394, 108 S.Ct. 1860 (Rehnquist, J., dissenting) (“[T]he question is whether a reasonable juror operating under the trial court’s instructions would have considered evidence of mitigating circumstances in a constitutional manner.”).
Our conclusion finds support in the Supreme Court’s decision in Penry v. Lynaugh, 492 U.S. 302, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989). In Penry, the Court considered a challenge to the Texas death penalty scheme, which limited the jury’s consideration of the appropriate sentence during a capital trial to resolving three “special issues.” Id. at 311-12, 109 S.Ct. 2934. The precise question before the Court was whether the defendant’s sentence was unconstitutional “because the jury was not adequately instructed to take into consideration all of his mitigating evidence and because the terms in the Texas special issues were not defined in such a way that the jury could consider and give effect to his mitigating evidence in answering them.” Id. at 313, 109 S.Ct. 2934. Despite Teague, the Court reversed the death sentence, finding that the jury instructions and special issues structure did not adequately allow the jury to consider the defendant’s mitigating evidence of mental retardation and an abused childhood. The Court stated that Lockett and Eddings, along with Jurek v. Texas, 428 U.S. 262, 96 S.Ct. 2950, 49 L.Ed.2d 929 (1976), “dictated” the rule that “Texas juries must, upon request, be given jury instructions that make it possible for them to give effect to th[e] mitigating evidence in determining whether the death penalty should be imposed.” Penry, 492 U.S. at 318-19, 109 S.Ct. 2934; see also Graham, 506 U.S. at 475, 113 S.Ct. 892 (indicating that the Lockett/Eddings line of cases prohibit situations in which “relevant mitigating evidence [is] placed beyond the effective reach of the sentencer,” and require that the jury must have a “reliable means of giving mitigating effect to” the mitigating evidence);12 Saffle, 494 U.S. at 491, 110 S.Ct. 1257 (same).
In arguing for the nonretroactivity of Mills, the Commonwealth invokes two oth*244er Supreme Court decisions handed down after Mills. First, the Commonwealth relies on the dissenting opinion and one of the concurring opinions in McKoy v. North Carolina, 494 U.S. 433, 110 S.Ct. 1227, 108 L.Ed.2d 369 (1990), to argue that reasonable minds could have disagreed on the outcome in Mills. See also Caspari, 510 U.S. at 395, 114 S.Ct. 948 (noting that conflicting holdings of state and federal courts indicates disagreement among reasonable minds); Sawyer v. Smith, 497 U.S. 227, 236-37, 110 S.Ct. 2822, 111 L.Ed.2d 193 (1990) (noting that the fact that three justices dissented in a prior ease casts doubt on the argument that the holding in that case was compelled by prior precedent); Miller v. Lockhart, 65 F.3d 676, 686 (8th Cir.1995) (citing opinions in McKoy). In McKoy, the Supreme Court affirmed Mills unconditionally and applied it to North Carolina’s sentencing scheme, holding that the unanimity scheme there, precisely like the one in Mills, unconstitutionally precluded jurors from giving effect to evidence they might believe called for a sentence less than death. McKoy, 494 U.S. at 439-440, 110 S.Ct. 1227. In the dissenting opinion and one of the concurring opinions, however, four justices in McKoy expressed some level of disagreement with the proposition that Mills was merely an application of the Lockett/Eddings rule. See id. at 452-457, 110 S.Ct. 1227 (Kennedy, J., concurring in the judgment); id. at 457-471, 110 S.Ct. 1227 (Scalia, J., dissenting). But see id. at 445-452, 110 S.Ct. 1227 (Blackmun, J., concurring) (stating that in Mills “[t]he Court concluded that a rule mandating unanimous agreement before any juror could consider a particular mitigating factor was forbidden by [its] decisions in Lockett and Eddings” (citations omitted)). Despite the Commonwealth’s suggestion otherwise, we are unconvinced that McKoy evidences anything but the correctness of our conclusion. The concurring and dissenting opinions in McKoy are not controlling authority, and our opinion today is in accord with the Court’s majority opinions in both Mills and McKoy, each of which independently makes clear that Mills was premised on a straight-forward application of settled constitutional precedent.13
Secondly, the Commonwealth places great reliance on the Supreme Court’s reasoning in Saffle v. Parks, 494 U.S. 484, 110 S.Ct. 1257, 108 L.Ed.2d 415 (1990). In Saffle, the Supreme Court upheld a jury instruction that required the jury to “avoid any influence of sympathy” in sentencing, finding that the rule sought by Saffle, which was grounded in the principles established in Lockett and Eddings, was a new rule under Teague. Id. at 487, 110 S.Ct. 1257. According to the Court, “Lockett and Eddings [did] not speak directly, if at all, to the issue presented” in *245Saffle, “whether the State may instruct the sentencer to render its decision on the evidence without sympathy.” Id. at 490, 110 S.Ct. 1257. In so concluding, the Court noted a distinction between rules relating “to what mitigating evidence the jury must be permitted to consider in making its sentencing decision,” and those relating “to hoiu it must consider the mitigating evidence.” Id. (emphasis in original). “There is a simple and logical difference,” the Court stated, “between rules that govern what factors the jury must be permitted to consider in making its sentencing decision and rules that govern how the State may guide the jury in considering and weighing those factors in reaching a decision.” Id. at 490, 110 S.Ct. 1257. An anti-sympathy instruction does not preclude the jury from considering any mitigating evidence; it merely instructs them to consider that evidence without recourse to sympathy. See id. at 493, 110 S.Ct. 1257 (“The State must not cut off full and fair consideration of mitigating evidence; but it need not grant the jury the choice to make the sentencing decision according to its own whims or caprice.”). Accordingly, the Court concluded that Lockett and Ed-dings do not prohibit such an anti-sympathy instruction. Id. at 490-94, 110 S.Ct. 1257; see also Brown, 479 U.S. at 543, 107 S.Ct. 837 (affirming constitutionality of an anti-sympathy instruction).
The Commonwealth, as well as our concurring colleague, seizes on Saffle’s “what” versus “how” distinction as supporting its view that Mills was a new rule; in effect, it argues that jury unanimity requirements made unconstitutional by Mills are nothing but a rule regarding how the jury must consider mitigating evidence, i.e., the jury must consider it unanimously. While perhaps viscerally appealing, upon reflection this linguistic shorthand does not withstand scrutiny. Indeed, we believe that, if anything, Saffle supports the conclusion that Mills did not announce a new rule. As discussed above, the harm identified in Mills by the Supreme Court — as well as the dissenting opinion in the Maryland Court of Appeals — was the potential that jurors were precluded from considering any and all mitigating evidence. See Mills, 486 U.S. at 375, 108 S.Ct. 1860. Quite unlike the anti-sympathy instruction considered in Saffle, Mills does concern the precise harm addressed by the Court in Lockett and Eddings: Its focus was on what — i.e., any and all — evidence jurors could consider. Although “unanimously” is clearly an adverb, the meaning of the term “unanimously” as used here does not actually relate to “how” the jury is to view the evidence (as does sympathy). Rather, here unanimity presents a barrier, potentially preventing jurors from considering any and all mitigating evidence. To the extent that Saffle suggests that such rules are clearly prohibited by Lockett and Eddings, it lends support to our conclusion here. See Saffle, 494 U.S. at 491, 110 S.Ct. 1257 (stating that preventing a jury from “considering, weighing, and giving effect to all of the mitigating evidence .... come[s] under the rule of Lockett and Eddings”).
Finally, we note that our conclusion finds additional support in a recent ruling by the Court of Appeals for the Sixth Circuit. In Gall v. Parker, 231 F.3d 265 (6th Cir.2001), the Sixth Circuit held that application of Mills on habeas was not prohibited by Teague, both because Mills did not announce a new rule and because Mills falls within the second of Teague’s exceptions to nonretroactivity. With regard to the former conclusion, which is more relevant here, the Sixth Circuit found that Mills was “dictated by the Lockett rule,” and emphasized the language in Mills quoted above, in which the Supreme Court made “clear[that] it did nothing more than apply Lockett to a new *246factual situation.” Id. at 328. The Sixth Circuit also noted the Supreme Court’s citation in Mills to Andres v. United States, 333 U.S. 740, 68 S.Ct. 880, 92 L.Ed. 1055 (1948), which the court characterized as “a half-century old death penalty reversal” in which “the Court granted a new trial after finding fault in instructions that ‘probab[ly]’ induced a ‘reasonable’ juror to conclude that unanimity was needed to ‘qualify’ a verdict of guilty in order to preclude a death sentence.” Gall, 231 F.3d at 323 (quoting Andres, 333 U.S. at 752, 68 S.Ct. 880). “Given Lockett, Andres, and the Court’s clear language in Mills,” the Sixth Circuit concluded that Mills did not impose a new constitutional obligation. Gall, 231 F.3d at 323.14
It is worth emphasizing here that Banks’s case is far stronger than the one presented to the Sixth Circuit in Gall. Whereas Gall’s conviction became final in 1981, limiting the legal landscape to primarily the plurality opinion in Lockett, Banks’ conviction was not final until late in 1987, only months before Mills and after the Supreme Court’s decisions in Eddings, Skipper, and Hitchcock, in all of which a clear majority of the Court applied the Lockett rule and reversed a sentence of death.
In contrast to the Sixth Circuit’s decision in Gall, two Courts of Appeals have held that Teague bars the application of Mills on habeas. In Miller v. Lockhart, 65 F.3d 676 (8th Cir.1995), the Eighth Circuit Court of Appeals held that Mills announced a new rule, but did so in a case in which, like Gall, only Lockett had been decided prior to the defendant’s conviction becoming final. Id. at 685. The question of whether Mills was dictated by the plurality opinion in Lockett alone may be a close one — as demonstrated by the contradictory opinions of the Sixth and Eighth Circuits — but the answer to the question before us is clear. When considering Lockett in conjunction with Eddings, in which a majority of the Court adopted the rule announced by the Lockett plurality, the result in Mills was obvious, especially given the Court’s continued reliance on and application of the Lockett rule, prior to Mills, in cases such as Skipper and Hitchcock.
The Court of Appeals for the Fifth Circuit has also repeatedly stated that Mills announced a new rule under Teague. See Woods v. Johnson, 75 F.3d 1017, 1036 (5th Cir.1996); Nethery v. Collins, 993 F.2d 1154, 1161-62 (5th Cir.1993); Wilcher v. Hargett, 978 F.2d 872, 877-78 (5th Cir.1992); Cordova v. Collins, 953 F.2d 167, 173 (5th Cir.1992). However, as we previously noted in Banks I, the court has not analyzed or explained its conclusions. Banks I, 271 F.3d at 542 n. 16. Accordingly, we find little in the Fifth Circuit’s decisions to persuade us that our application of Mills is prohibited by Teague.15
*247TV.
Our previous ruling in this case was reversed by the Supreme Court only insofar as we held it unnecessary to decide whether Mills had retroactive application. Because we now hold that our application of Mills on habeas review of Banks’s sentence was not prohibited by Teague, we do not disturb the remainder of our previous opinion, including its discussion and holding with regard to the merits of Banks’s Mills claim. We merely augment that opinion by essentially replacing its discussion of the Teague issue with the analysis here. Accordingly, our judgment requiring a new penalty phase for Banks will remain unchanged.
. A conviction becomes final for Teague purposes "when the availability of direct appeal to the state courts has been exhausted and the time for filing a petition for a writ of certiora-*231ri has elapsed or a timely filed petition has been finally denied.” Caspari v. Bohlen, 510 U.S. 383, 390, 114 S.Ct. 948, 127 L.Ed.2d 236 (1994). Banks’s conviction was therefore final when the Supreme Court denied certiorari on October 5, 1987. Banks v. Pennsylvania, 484 U.S. 873, 108 S.Ct. 211, 98 L.Ed.2d 162 (1987).
. On July 12, 2002, Banks filed a Petition for Rehearing. The Supreme Court recalled its issuance of judgment on July 17, but on August 26, the Court denied rehearing and reissued judgment.
. This is not the first time we have been presented with the issue of the retroactivity of Mills. In Zettlemoyer v. Fulcomer, 923 F.2d 284, 306 n. 19 (3d Cir.1991), we "decided to reach the merits of the Mills claim ... [but] did not expressly hold whether Mills falls outside the Teague bar.” Frey v. Fulcomer, 132 F.3d 916, 920 n. 4 (3d Cir.1997). In Frey, we noted the issue but did not reach it because the Commonwealth failed to raise it, and we deemed it waived. Id.
.The Maryland Court of Appeals upheld the sentence of death. See Mills v. State, 310 Md. 33, 527 A.2d 3 (1987). Although it agreed that "if the statute and form were read as petitioner suggested, jurors would be improperly prevented from giving due consideration to mitigating evidence,” see Mills, 486 U.S. at 372, 108 S.Ct. 1860 (emphasis in original), the Court of Appeals adopted a construction of the statute exonerating it from the potential constitutional issue. See Mills, 527 A.2d at 12-17.
.Because we conclude that Mills did not announce a new rule under Teague, we need not address Banks's arguments regarding whether Mills falls within one of the two exceptions to nonretroactivity. We note, however, that in Williams v. Dixon, 961 F.2d 448, 454-56 (4th Cir.1992), the Court of Appeals for the Fourth Circuit held that Teague does not bar application of Mills (and McKoy v. North Carolina, 494 U.S. 433, 110 S.Ct. 1227, 108 L.Ed.2d 369 (1990)) on habeas because they are "watershed rules of criminal procedure implicating the fundamental fairness and accuracy of the criminal proceeding.” Graham, 506 U.S. at 478, 113 S.Ct. 892 (quotations omitted); see also Gall v. Parker, 231 F.3d 265, 323 (6th Cir.2000) (holding that Mills did not announce a new rule under Teague but also finding that, even if Mills did announce a new rule, it falls within the second Teague exception); Jermyn v. Horn, No. 97-634, 1998 WL 754567, at *36-*39 (M.D.Pa. Oct. 27, 1998) (holding that Mills is a new rule but falls within the second Teague exception).
. It is worth noting that although the Supreme Court has instructed that "the reasonable views of state courts are entitled to consideration” as part of the legal landscape, Caspari v. Bohlen, 510 U.S. 383, 395, 114 S.Ct. 948, 127 L.Ed.2d 236 (1994), both parties have focused exclusively on Supreme Court precedent. At any rate, we believe that here the Supreme Court's case law adequately depicts the legal landscape at the time of Banks's conviction.
. The Court's opinion in Woodson was a plurality of three — Justices Powell, Stevens, and Stewart. Justices Brennan and Marshall both concurred in the judgment given their opinion *236that capital punishment inherently violates the Eighth and Fourteenth Amendments.
. Lockett's plurality opinion was written by Chief Justice Burger and joined by Justices Powell, Stevens, and Stewart. Justice Brennan took no part in the case, and Justice Marshall again concurred on the ground that capital punishment is always unconstitutional.
. The Supreme Court has made clear that Brown was not dictated by Lockett and Eddings. See Saffle, 494 U.S. at 494, 110 S.Ct. 1257. Nonetheless, the decision constitutes part of the legal landscape prior to Banks’s conviction becoming final.
. In her concurrence, Judge Sloviter focuses on whether Mills was "commanded” or "dictated” by the legal landscape in concluding that it announced a new rule.
. Although it has been noted that for Teague purposes an opinion's language is not always "conclusive,” see, e.g., Butler, 494 U.S. at 415, 110 S.Ct. 1212, we nonetheless find the Court's rhetoric and reasoning in Mills to be unique — providing extensive "evidence tending to prove” that Mills did not announce a new rule. O’Dell, 521 U.S. at 161 n. 2, 117 S.Ct. 1969.
. Unlike our concurring colleague, we do not view Graham v. Collins as adding anything of significance to the legal landscape. In Graham, the Supreme Court decided that, unlike Penry — in which the Court held that the "special issues” structure did not give the jury a genuine opportunity for consideration of diminished capacity attributes such as mental retardation and an abused childhood — the “special issues” structure did place certain mitigating evidence such as youth, family background and positive character traits within the sentencer’s "effective reach.” Graham, 506 U.S. at 475, 113 S.Ct. 892. Thus, a result in Graham's favor was not dictated by Penry and the Penry rule could not be given retroactive effect.
. Moreover, to the extent the Commonwealth relies on the McKoy minority's view that Mills was not actually decided on Lockett/Eddings grounds, we are unpersuaded that their argument is particularly relevant. Asking whether Mills is retroactive in this case is shorthand for asking whether the rule Banks seeks to have applied on habeas — that the Constitution prohibits unanimity instructions that preclude jurors from giving proper consideration to proffered mitigating evidence— was a new rule under Teague given the date his conviction became final. It is for similar reasons that, as we discuss below, our decision here is not formally inconsistent with, for instance, the Eighth Circuit's decision in Miller v. Lockhart, 65 F.3d 676 (8th Cir.1995), that Mills was a new rule. Our conclusion that Mills may be retroactively applied on habeas is essentially an acknowledgment that what Banks seeks (and what Mills sought) is a clear application of the Lockett/Eddings rule, a constitutional principle well settled prior to his conviction becoming final. Thus, it could be argued that it is essentially irrelevant whether or not the rule under inquiry was actually decided in Mills.
. Similarly, in an opinion we have previously referred to as "brief but thoughtful,” Frey v. Fulcomer, 132 F.3d 916, 920 n. 4 (3d Cir.1997), the District of Delaware concluded in 1993 that the "requirement that juries in capital cases be permitted to consider all mitigating factors and aspects of a defendant’s character and to give effect to that evidence was firmly established” in Lockett and Eddings. DeShields v. Snyder, 829 F.Supp. 676, 688 (D.Del.1993) (quoting Byrd v. Delo, 733 F.Supp. 1334 (E.D.Mo.1990)). Accordingly, the court found that Mills was “nothing more than a mere extension of then existing precedent to a new factual scenario.” Id.; see also, e.g., Hopkinson v. Shillinger, 781 F.Supp. 737, 742 (D.Wy.1991) ("The Eighth Amendment [under Lockett and Eddings] prohibits barriers to consideration of mitigating evidence whether they result from evidentiary rulings, statute, or jury instructions. Thus, Mills and McKoy are simply different factual applications of that established principle and are applicable [on habeas].”).
. We also note that in McDougall v. Dixon, 921 F.2d 518, 539 (4th Cir.1990), the Court of *247Appeals for the Fourth Circuit indicated in dicta that Mills and McKoy were new rules under Teague. In Williams v. Dixon, 961 F.2d 448, 453 n. 3 (4th Cir.1992), however, the court found it unnecessary to consider whether Mills and McKoy were in fact new rules because it held that they fell within the second Teague exception.