OPINION
SMITH, Circuit Judge.Eric Greene petitioned for relief under 28 U.S.C. § 2254 from his state court convictions for second degree murder, robbery, and conspiracy. This appeal requires us to resolve the thorny question of what the temporal cutoff is for determining “clearly established Federal law” for purposes of the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”) standard of review, set forth in 28 U.S.C. § 2254(d)(1). Based on the statute’s text and Supreme Court precedent, we now hold that “clearly established Federal law” should be determined as of the date of the relevant state-court decision. Because the Supreme Court decision that Greene wishes to rely upon in his habeas petition, Gray v. Maryland, 523 U.S. 185, 118 S.Ct. 1151, 140 L.Ed.2d 294 (1998), had not yet been decided at the time of the relevant state-court decision, he cannot show that his state court proceedings resulted in an unreasonable application of “clearly established Federal law, as determined by the Supreme Court of the United States[.]” 28 U.S.C. § 2254(d)(1). Thus, we will affirm the judgment of the District Court denying Greene’s habeas petition.
I.
The Crime
In early December of 1993, three or four men robbed a small family owned grocery store in North Philadelphia, and its owner, *88Francisco Azcona, died after being shot at point-blank range. When the robbers were unable to open the cash register, they picked it up and carried it out of the store, escaping in a station wagon parked nearby. A week after Mr. Azcona’s murder, Greene, Julius Jenkins, Atil Finney, and Gregory Womack robbed a check cashing facility. They were apprehended shortly thereafter and the police seized a firearm. Through ballistics testing, the police determined that the seized firearm was used in Mr. Azcona’s murder. With this evidence, the police were able to make progress in the murder investigation.
The Investigation
In late February of 1995, Detective Robert Snell of the Philadelphia Police Department questioned Demond Jackson about his involvement in Mr. Azcona’s murder. Jackson admitted that he was in the station wagon parked outside of the grocery store the night of the murder, but he claimed that he was simply getting a ride to West Philadelphia when the others stopped at the store. He described how several of them went inside and committed the murder. He also identified Jenkins as the shooter, and indicated that Naree Abdullah carried the cash register out of the store. According to Jackson, Finney entered the store with Jenkins and Abdullah, while Greene remained with the driver and Jackson in the automobile. Jackson added that the proceeds of the robbery were split among only five of the men, and that he did not take a share of the proceeds.
Armed with the information from Jackson, Detective Michael Gross of the Philadelphia Police Department questioned Finney in early March of 1995. Finney gave a statement to the police in which he admitted that he was one of the participants in the grocery store robbery and that he was inside the store at the time of the robbery. He identified Jenkins as the shooter, implicated Greene as the robber who carried the cash register out of the store, indicated that Womack had driven the station wagon, and stated that another individual was involved in the robbery. Although Finney initially stated that five people were involved in the robbery, he later noted that there were six people in the car. A few days later, Detective Joseph Walsh of the Philadelphia Police Department questioned Womack. Womack gave a statement to the police in which he admitted he was the driver of the station wagon the night that Jenkins killed Mr. Azcona. In addition, he implicated Finney, Abdullah, and Greene in the robbery.
Shortly after these statements were obtained, first degree murder charges were filed against Jenkins. Greene, Finney, Womack, and Abdullah were charged with, inter alia, second degree murder, three counts of robbery, and conspiracy.
The Trial
Greene filed a pretrial motion seeking severance on several grounds. In that motion, he argued, inter alia, that a joint trial with his codefendants would be prejudicial because of the incriminating statements they had made to authorities. As support for his motion, Greene cited Bruton v. United States, 391 U.S. 123, 88 S.Ct. 1620, 20 L.Ed.2d 476 (1968), and Richardson v. Marsh, 481 U.S. 200, 107 S.Ct. 1702, 95 L.Ed.2d 176 (1987). During a pretrial hearing, Greene urged the trial court, the Court of Common Pleas of Philadelphia, to sever the trials because the statements of some of his non-testifying codefendants implicated him and identified him as the person who carried the cash register out of the grocery store. The trial court, recognizing that the statements might be inadmissible at a joint trial, but also noting that redaction might resolve any problem *89of prejudice, posed a hypothetical to the parties:
Judge: It’s unusual to do this, but suppose the statement is redacted to say, “I didn’t take the cash register.” In other words, each defendant’s statement would state—
[Greene’s Counsel]: I didn’t take it.
Judge: I didn’t. Someone else took the cash register.
[Greene’s Counsel]: There’s another nuance that I want Your Honor to know because your suggestion is brilliant. I just want to factor in one more thing. Of the three co-defendants who gave statements, indeed, Atil Finney says that my client took the cash register and Mr. Gregory Womack made the statement that my client was involved in that, but believe it or not, Julius Jenkins says that someone entirely different — he says that either Naree or another person ... took the cash register, either Womack or Naree, so you have Mr. Jenkins saying specifically that one of two other people took it — people involved, I might add, but not [Greene].
In response, the Court declared that the statements were not interlocking and inquired how the conflicting statements “pin point[ed]” Greene since the jury would have “information that three different people have been named as having taken the cash register.” Greene’s counsel replied that the Court’s analysis was “excellent,” but wondered how the Commonwealth would redact the statements. The Court replied that “it seems to me that the fair way to redact these [statements] is to refer to three different people.” Greene’s counsel responded: “As long as I would be allowed to argue in my closing speech that you heard what you heard and you heard that there were different people, then I would have no problem with [it].” The prosecutor offered to redact the statements so that “not one specific person carries out the cash register.” Greene’s counsel agreed that, under Bruton, such a redaction would remove any prejudice from the statements. Greene’s counsel also pressed, without success, an additional basis for severance: that Jenkins, who would be tried alongside Greene, was facing a capital murder charge.
Greene’s trial was held in February of 1996. At trial, Mr. Azcona’s wife and sister-in-law identified Jenkins as the shooter, but they were unable to identify any of the other robbers in the store or to state with certainty whether there were three or four men involved in the robbery.
Jackson, who had not been charged with any crimes associated with the robbery, was the prosecution’s star witness. His testimony differed significantly from his statement to Detective Snell. Jackson testified that all of the occupants of the station wagon went into the store except for him and Womack. Contrary to his earlier statement that Abdullah picked up the cash register, Jackson testified that Greene took the cash register from the store. Jackson was cross-examined extensively on the differences between his earlier statement and his trial testimony. In addition, Greene attacked Jackson’s credibility by highlighting that Jackson was present in Womack’s station wagon but had not been charged with any crimes related to the robbery and that Jackson had outstanding drug charges.
The Commonwealth also called Detectives Gross and Walsh to testify about the statements they obtained from Finney and Womack. Neither Greene nor his codefendants objected to the reading of those statements in redacted form. Detective Gross read Finney’s redacted statement, which substituted the nicknames or proper names of Finney’s codefendants with the *90phrases “this guy,” “other guys,” and “two guys.” The redacted statement also used the neutral pronouns “we” or “someone” in certain instances. For example, Finney’s initial description of the robbery in its redacted form read:
We were riding around in this, this guy’s car, me and three other guys were in the North Philadelphia. We — when one said let’s get paid. Everyone said okay and we saw this store. So me and two guys went in the store. When we got inside two guys stayed up front and I stayed to the back. One guy had his gun on the guy and was at the cash register getting the money, but it wouldn’t open. I heard a shot and looked over. Blood was coming out of the guy’s mouth. After that someone grabbed the register and we all ran out.
In at least one instance, when Detective Gross reached a portion of the statement where Finney was asked to identify “these guys” by their full names, the redaction simply deleted the names. The redacted answer stated: “One is and the other is.” Later in the statement, when Finney was asked if he recognized anyone in certain photographs, the redacted answer stated: “Number three is. Number six is. Number eight is.” The redacted statement also replaced the names of certain defendants with the word “blank” on.four occasions. During Detective Gross’s reading of the redacted statement, the trial court instructed the jury that Finney’s statement could only be considered as evidence against him and not as evidence against any other defendant.
Detective Walsh, during his testimony, read a redacted version of Womack’s statement. In it, Womack declared:
It was me and another guy. We were in the car, the other three went in the store. The car was around the corner and then they came out of the store
carrying a cash register. As we pulled off the shooter said he shot the guy. I think someone asked him why did he shoot the guy? He didn’t answer him. Then we drove up our neighborhood. Then we drove to someone’s and we opened the register and got the money and we took the register and dumped it in a trash dumpster in a townhouse.
Although the redacted statement utilized neutral references such as “guy,” “another guy,” “someone,” “someone else,” “one,” and “others,” it replaced the names of some of the codefendants with the word “blank” on three occasions. The trial court did not give a limiting instruction following the reading of Womack’s redacted statement, and neither Greene nor any of his codefendants requested such an instruction.
After closing arguments, the trial court issued a limiting instruction directing the jurors not to consider either redacted statement as evidence against any defendant other than the declarant. The jury found Greene guilty of second degree murder, three counts of robbery, and one count of conspiracy. The trial court sentenced him to life imprisonment.
Subsequent Procedural History
Greene filed a direct appeal with the Pennsylvania Superior Court. Citing Bruton, Greene argued that his trial should have been severed from that of his codefendants because the statements implicating him “were not suitable for redaction.” On December 16, 1997, the Pennsylvania Superior Court affirmed the judgment against Greene, addressing his Bruton claim on the merits. The Court observed that the statements that were admitted into evidence “were redacted to remove any reference to the other defendants in the case” and “[t]he trial court instructed the jury on more than one occasion that *91such statements could only be considered as evidence against the defendants who made them.” In light of these observations, the Pennsylvania Superior Court concluded that Bruton was not violated and that Greene was not deprived of his right to confrontation.
Greene filed a timely petition for allowance of appeal with the Pennsylvania Supreme Court. His petition argued, inter alia, that he had been deprived of his rights under the Confrontation Clause by the introduction of Womack’s and Finney’s statements. As support for his position, Greene again cited Bruton. While Greene’s petition for allowance of appeal was pending with the Pennsylvania Supreme Court, the United States Supreme Court issued its decision in Gray. In Gray, the Supreme Court stated that “considered as a class, redactions that replace a proper name with an obvious blank, the word ‘delete,’ a symbol, or similarly notify the jury that a name has been deleted are similar enough to Bruton’s unredacted confessions as to warrant the same legal results.” 523 U.S. at 195, 118 S.Ct. 1151. Thereafter, the Pennsylvania Supreme Court granted Greene’s petition for allocatur “limited to the issue of whether the common pleas court erred by denying the motion for severance thereby resulting in the violation of [Greene]’s Sixth Amendment right of confrontation upon the admission of statements given by his nontestifying codefendants.” Commonwealth v. Trice, 552 Pa. 201, 713 A.2d 1144 (1998). After granting the petition for allocatur, however, the Pennsylvania Supreme Court dismissed Greene’s appeal as improvidently granted. Commonwealth v. Trice, 556 Pa. 265, 727 A.2d 1113 (1999). Greene’s conviction became final ninety days later, on July 28, 1999, when the time period for filing a petition for certiorari to the United States Supreme Court expired. See Kapral v. United States, 166 F.3d 565, 570-71 (3d Cir.1999).
In early August of 1999, Greene sought relief from his conviction based on Pennsylvania’s Post Conviction Relief Act (“PCRA”), 42 Pa. Cons.Stat. §§ 9541-9546. In his PCRA petition, Greene argued that the trial court abused its discretion in denying the severance motion, and cited, inter alia, the prosecutor’s summation, which allegedly improperly informed the jury that Finney’s statement corroborated that the others on trial were implicated in the commission of the crime. The PCRA petition did not assert a Confrontation Clause claim as it failed to reference the redacted statements or to cite the Supreme Court’s decisions in Bruton, Marsh, or Gray. The trial court dismissed Greene’s PCRA petition as frivolous. Greene, acting pro se, appealed the denial of his PCRA petition to the Pennsylvania Superior Court, asserting that the trial court erred by refusing to grant a severance. His argument cited only Pennsylvania authority regarding motions to sever multiple criminal charges. He did not refer to the Confrontation Clause, Bruton, Marsh or Gruy. On December 31, 2003, the Pennsylvania Superior Court affirmed the dismissal of Greene’s PCRA petition, noting that the severance claim had been finally litigated and could not afford him collateral relief. Greene filed another petition for allowance of appeal with the Pennsylvania Supreme Court, which denied allocatur.
This timely § 2254 petition followed. In his petition, Greene asserted, inter alia, that his trial should have been severed “due to antagonistic defenses, due to the fact a codefendant was subjected to the death penalty even though petitioner was not, and particularly due to the fact that effective redaction of the codefendant’s [sic] statements, though attempted, was *92polluted by gross prosecutorial misconduct.” In a comprehensive report, the Magistrate Judge to whom the petition had been referred recommended that Greene’s petition be dismissed, but that a certificate of appealability be granted on the Confrontation Clause claim arising out of the introduction of Womack’s and Finney’s redacted statements at trial.
The Magistrate Judge struggled with whether to determine the “clearly established Federal law” under § 2254(d)(1) as of the date of the relevant state-court decision, as instructed by Justice O’Connor in her portion of the majority decision in Williams v. Taylor, 529 U.S. 362, 403-412, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000), or by the date Greene’s conviction became final, as instructed by Justice Stevens in his portion of the majority decision in Williams, id. at 390, 120 S.Ct. 1495. This issue was significant because it determined whether “clearly established Federal law” for purposes of Greene’s § 2254 petition included the Supreme Court’s decision in Gray. If the cutoff date was the date of the relevant state-court decision, i.e., the Pennsylvania Superior Court’s December 16, 1997 decision affirming Greene’s convictions on direct appeal, that date preceded the Supreme Court’s decision in Gray, and Gray would not be part of the “clearly established Federal law” applicable to this habeas petition.1 But if the date Greene’s convictions became final, July 28, 1999, was the pertinent cutoff date, Gray, which was issued more than a year earlier on March 9, 1998, would be “clearly established Federal law.”
The Magistrate Judge ultimately determined that the controlling date for ascertaining the “clearly established Federal law” for Greene’s habeas petition was the date of the relevant state-court decision. Accordingly, the Magistrate Judge applied the Supreme Court law existing at the time of the Pennsylvania Superior Court’s December 16, 1997 decision, Bruton and Marsh, to determine whether Greene’s § 2254 petition merited relief. He concluded that the Pennsylvania Superior Court did not unreasonably apply Bruton and Marsh in concluding that the redacted statements did not violate the Confrontation Clause and recommended that the District Court deny the § 2254 petition.
The Commonwealth objected to the Magistrate Judge’s report, arguing that Greene had not procedurally exhausted his Confrontation Clause claim. The District Court overruled the Commonwealth’s objections, noting that Greene presented a general claim regarding the redacted con*93fessions and relied upon relevant Supreme Court authority, Bruton and Marsh. The District Court adopted the Magistrate Judge’s report and recommendation. The Court denied the petition, but also granted a certificate of appealability limited to Greene’s Confrontation Clause claim.2
II.
Before us, the Commonwealth argues that Greene did not satisfy the fair presentation requirement because he consented to the redactions in the trial court and did not fairly present his Confrontation Clause argument to the Pennsylvania Superior Court.3 It submits that Greene only raised the Confrontation Clause argument in his direct appeal to the Pennsylvania Supreme Court. We disagree.
The fair presentation requirement arises from the exhaustion doctrine pertaining to federal habeas review of state court decisions. Picard, v. Connor, 404 U.S. 270, 275, 92 S.Ct. 509, 30 L.Ed.2d 438 (1971). Fair presentation requires giving the state court the “first opportunity to hear the claim sought to be vindicated in [the] federal habeas proceeding[.]” Id. at 276, 92 S.Ct. 509. In this first opportunity, the habeas petitioner need merely “present [the] federal claim’s factual and legal substance to the state courts in a manner that puts them on notice that a federal claim is being asserted.” McCandless v. Vaughn, 172 F.3d 255, 261 (3d Cir.1999). In other words, the claims raised in the state courts must be substantially equivalent to the claim pressed in the federal court. Doctor v. Walters, 96 F.3d 675, 678 (3d Cir.1996), abrogated on other grounds by Beard v. Kindler, — U.S. -, 130 S.Ct. 612, — L.Ed.2d - (2009).
Greene fairly presented the factual and legal substance of his Confrontation Clause claim to the Pennsylvania state courts. On direct appeal, Greene presented his Confrontation Clause claim and the Pennsylvania Superior Court addressed the merits of that claim on the “basis of its substance, rather than on a procedural, or other ground.” Thomas v. Horn, 570 F.3d 105, 115 (3d Cir.2009) (citations omitted). Thus, Greene’s direct appeal satisfied the fair presentation requirement. Picard, 404 U.S. at 275, 92 S.Ct. 509.
III.
Having determined that Greene fairly presented his Confrontation Clause claim in the Pennsylvania state courts, we turn to a vexing issue that has, for the most part, evaded analytical discussion by the Supreme Court and the Courts of Appeals.4 That is, whether “clearly estab*94lished Federal law” under 28 U.S.C. § 2254(d)(1) is determined based on the “time of the relevant state-court decision,” Williams, 529 U.S. at 412, 120 S.Ct. 1495 (O’Connor, J., for the Court), the “time [the] state-court conviction became final,” id. at 390, 120 S.Ct. 1495 (Stevens, J., for the Court), or some combination thereof, e.g., Horn v. Banks, 536 U.S. 266, 272, 122 S.Ct. 2147, 153 L.Ed.2d 301 (2002) (per curiam) (holding that “in addition to performing any analysis required by AEDPA, a federal court considering a habeas petition must conduct a threshold Teague [v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989),] analysis when the issue is properly raised by the state”). The Supreme Court, until recently, appeared to have settled on the date of the relevant state-court decision.5 But the use of the date the petitioner’s conviction became final has refused to quietly exit the stage. In recent months, the Supreme Court has noted the “uncertainty” surrounding the meaning of “clearly established Federal law” for the purposes of § 2254(d)(1). Smith v. Spisak, — U.S. -, 130 S.Ct. 676, 681, — L.Ed.2d-(2010) (assuming that Supreme Court decision that was issued after the state supreme court affirmed petitioner’s conviction, but before the date petitioner’s conviction became final, constituted “clearly established Federal law”); see id. at 689 n. 2, 130 S.Ct. 676 (Stevens, J., concurring) (supporting date conviction became final); see also Thaler v. Haynes, — U.S. -, 130 S.Ct. 1171, 1174 n. 2, — L.Ed.2d - (2010) (per curiam) (noting in dicta that a certain case could not have constituted “clearly established Federal law” because the ease was “decided ... nearly six years after [petitioner’s] conviction became final and more than six years after the relevant state-court decision”); see Brown v. Greiner, 409 F.3d 523, 533 n. 3 (2d Cir.2005) (noting uncertainty caused by “inconsistent guidance” from the Supreme Court); Newland v. Hall, 527 F.3d 1162, 1198 n. 62 (11th Cir.2008) (noting uncertainty); but see Foxworth v. St. Amand, 570 F.3d 414, 430-32 (1st Cir.2009) (declaring relevant state-court decision approach “untenable” and stating that Supreme Court precedent leads to the “inexorable conclusion” that the date the conviction became final is the correct approach).6
After careful consideration of the divergent approaches to determining what constitutes “clearly established Federal law” *95under § 2254(d)(1), we now hold that the date of the relevant state-court decision is the controlling date. After surveying the questions that arise from the Supreme Court’s Williams decision, and considering the statutory text and post-WiMmms Supreme Court precedent, our view is that using the date of the relevant state-court decision to determine “clearly established Federal law” is the most logical approach to applying § 2254(d)(1).
A.
It is understandable that confusion surrounds what constitutes “clearly established Federal law.” In discussing the meaning of the AEDPA amendments, the Supreme Court has held that the “statutory phrase [‘clearly established Federal law’] referred] to the holdings, as opposed to the dicta, of [its] decisions as of the time of the relevant state-court decision.” Williams, 529 U.S. at 412, 120 S.Ct. 1495 (O’Connor, J., for the Court) (emphasis added). It has also held that all Supreme Court jurisprudence that would “qualify as an old rule under [its] Teague jurisprudence w[ould] constitute ‘clearly established Federal law ... ’ under § 2254(d)(1).” Id.
These statements from Justice O’Connor present the first area of confusion in Williams. The most logical meaning for the term “old rule,” a term that lacks any meaningful discussion post -Williams, is any rule which is not “new” under Teague. See Teague, 489 U.S. at 300-01, 109 S.Ct. 1060 (plurality opinion). If that is the case, then an “old rule” is any rule that was “dictated by the governing precedent existing at the time when [the petitioner’s] conviction became final[.]” Whorton v. Bockting, 549 U.S. 406, 417, 127 S.Ct. 1173, 167 L.Ed.2d 1 (2007) (defining new rule). In that event, the inclusion of old rules under Teague as “clearly established Federal law” would include Supreme Court decisions issued after the relevant state-court decision but before the petitioner’s conviction became final. Such an outcome, in our view, contradicts Justice O’Connor’s initial declaration that “clearly established Federal law” should be determined based on the date of the relevant state-court decision.7
*96To farther complicate Williams, the Supreme Court also held that the “threshold question under AEDPA is whether [a petitioner] seeks to apply a rule of law that was clearly established at the time his state-court conviction became final.” Williams, 529 U.S. at 390, 120 S.Ct. 1495 (Stevens, J., for the Court) (emphasis add*97ed).8 Thus, the majority opinions of the Court, on their faces, offered differing interpretations of the “clearly established Federal law” language.
Supreme Court precedent after Williams has also raised questions. At least some post-Williams authority suggests that the Teague test and § 2254(d)(1) are distinct inquiries. E.g., Horn, 536 U.S. at 272, 122 S.Ct. 2147; see Thaler, 130 S.Ct. at 1174 n. 2. The instances where both tests must be met, however, are unclear.9 More importantly, it is also unclear whether the distinct nature of the two inquiries has any impact on how we approach the meaning of “clearly established Federal law” for the purposes of § 2254(d)(1).
In sum, we have (1) Justice O’Connor’s majority opinion in Williams, which seems to contradict itself by stating that the date of the relevant state-court decision is the operative date for determining “clearly established Federal law” while simultaneously stating that Supreme Court jurisprudence that would qualify as “old rules” under Teague (which relies on the date the petitioner’s conviction became final) is also “clearly established Federal law,” (2) Justice Stevens’s majority opinion in Williams, which contradicts Justice O’Con-nor’s directive that we should look to the date of the relevant state-court decision, and (3) post -Williams Supreme Court authority suggesting that Teague and § 2254(d)(1) are distinct inquiries subject to independent analysis under certain circumstances.
While many courts have managed to avoid confronting these issues, e.g., Spisak, 130 S.Ct. at 681, this case presents us with the inescapable obligation to decide the cutoff date for determining “clearly established Federal law.” Greene’s petition turns on whether he may invoke Gray; without that decision he cannot obtain relief. See infra Section IV. Gray was decided on March 9, 1998. Thus, using the date of the relevant state-court decision, the Pennsylvania Superior Court’s December 16, 1997 decision, Gray would not be “clearly established Federal law.” But using the date Greene’s conviction became final, July 28, 1999, Gray would be “clearly established Federal law.” Indeed, Greene’s case is the perfect storm of facts for resolving the issue of which date — the date of the relevant state-court decision or the date the state-court conviction became final — should be used for determining “clearly established Federal law” for the purposes of § 2254(d)(1).
B.
The text of § 2254(d)(1) supports using the date of the relevant state-court decision for determining “clearly established Federal law.” Section 2254(d)(1) is concerned with “decision[s]” that were “contrary to” or “unreasonable application^]” of “clearly established Federal law”:
An application for a writ of habeas corpus on behalf of a person in custody pursuant to the judgment of a State court shall not be granted with respect to any claim that was adjudicated on the merits in State court proceedings unless the adjudication of the claim' — •
(1) resulted in a decision that was contrary to, or involved an unreasonable *98application of, clearly established Federal law, as determined by the Supreme Court of the United States[.]
Id. The statute indicates that a “decision” results from a state court’s adjudication “on the merits” of a claim. Id. In other words, the decision occurs when the state court has acted on the substance of a petitioner’s claim. See Thomas, 570 F.3d at 115 (concluding that “adjudicated on the merits” means that the state ruling resolved the claim “on the basis of its substance, rather than on a procedural, or other ground”). Thus, it is the state court’s resolution of the petitioner’s claim that must be “contrary to” or an “unreasonable application” of existing Federal law to justify granting habeas relief. 28 U.S.C. § 2254(d)(1); see Newland, 527 F.3d at 1199 (holding that “the highest state court decision reaching the merits of a habeas petitioner’s claim is the relevant state court decision”).
Given that AEDPA is concerned with the review of the state court’s decision on the merits of the petitioner’s claim, the statute, read in the most straightforward fashion, requires that the relevant Federal law be “clearly established” at the time of that state-court decision. 28 U.S.C. § 2254(d)(1).10 Reading the language plainly, “clearly established” contemplates that the law or precedent existed at the time of the state court’s substantive resolution of the petitioner’s claim. Cf. United States v. Atiyeh, 402 F.3d 354, 364 (3d Cir.2005) (“Congress kn[o]w[s] how to use the past tense.”). A state court cannot unreasonably apply a Supreme Court decision that did not exist at the time of its decision. See Priester v. Vaughn, 382 F.3d 394, 400 (3d Cir.2004) (noting that the Pennsylvania Superior Court “did not act unreasonably in failing to predict the Supreme Court’s decision in Gray”). The same is true for the “contrary to” prong of the statute.
C.
Supreme Court decisions after Williams further bolster our conclusion. In Lockyer v. Andrade, 538 U.S. 63, 123 S.Ct. 1166, 155 L.Ed.2d 144 (2003), the Supreme Court stated unequivocally that “ ‘clearly established Federal Law” under § 2254(d)(1) is the governing legal principle or principles set forth by the Supreme Court at the time the state court renders its decision.” Id. at 71-72, 123 S.Ct. 1166. The same test was repeated in Wiggins v. Smith, 539 U.S. 510, 520, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003), Yarborough v. Alvarado, 541 U.S. 652, 660-61, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004), and Carey v. Musladin, 549 U.S. 70, 74, 127 S.Ct. 649, 166 L.Ed.2d 482 (2006). Accord 1-2 Randy Hertz & James S. Liebman, Federal Habeas Corpus Practice and Procedure, Fifth Edition § 2.4 (stating that the date for determining “clearly established Federal law” for § 2254 petitions is “the date of the state court decision”); 2-32 id. § 32.3 (“Section 2254(d)(1) limits federal review to legal rules that actually were in effect when the state court decided the case.”).
The date the conviction became final, on the other hand, has not gained much traction in the Supreme Court. Aside from stating that Teague and § 2254(d)(1) are distinct inquiries, Horn, 536 U.S. at 272, 122 S.Ct. 2147, and that in certain circumstances both Teague and § 2254(d)(1) must be satisfied, id., the Supreme Court has not suggested that the date the conviction *99became final has any import in determining “clearly established Federal law” for the purposes of § 2254(d)(1). In fact, it appears that Justice Stevens’s majority opinion language from Williams stating that the “threshold question” is whether the petitioner seeks to apply a rule that was clearly established at the time his state-court conviction became final, Williams, 529 U.S. at 390, 120 S.Ct. 1495, has been supplanted by Lockyer, where the Supreme Court agreed that the “threshold matter” was to decide what constituted “clearly established Federal law,” but then used the relevant state-court decision date to determine that law, 538 U.S. at 71, 123 S.Ct. 1166. The most telling observation regarding the use of the date the conviction became final is that the strongest authorities we have found for that approach are the recent Supreme Court opinions expressing uncertainty on which date is appropriate. See Spisak, 130 S.Ct. at 681; see also Thaler, 130 S.Ct. at 1174 n. 2. Mere uncertainty cannot counterbalance the numerous Supreme Court decisions that have unequivocally, albeit without analysis, taken the other approach. As an inferior federal court, we are not free to ignore the numerosity of these pronouncements.
Moreover, it appears that Justice Stevens’s primary concern with Justice O’Connor’s formulation of the “clearly established Federal law” inquiry is her view that the phrase “refers to the holdings, as opposed to the dicta, of [the Supreme Court’s] decisions[.]” Williams, 529 U.S. at 412, 120 S.Ct. 1495 (O’Connor, J., for the Court). In Carey, Justice Stevens explained that he took issue with Justice O’Connor’s formulation because it discouraged state courts from seeking guidance from the Supreme Court’s decision on the grounds that such guidance was dicta:
Virtually every one of the Court’s opinions announcing a new application of a constitutional principle contains some explanatory language that is intended to provide guidance to lawyers and judges in future cases. It is quite wrong to invite state court judges to discount the importance of such guidance on the ground that it may not have been strictly necessary as an explanation of the Court’s specific holding in the case. The text of [AEDPA] itself provides sufficient obstacles to obtaining habeas relief without placing a judicial thumb on the warden’s side of the scales.
Carey, 549 U.S. at 79, 127 S.Ct. 649 (Stevens, J., concurring) (citations omitted). This concern exists independent of the date upon which “clearly established Federal law” is determined and is not implicated in the issue we decide today. The decisions preceding Gray — Bruton and Marsh — explicitly refused to provide guidance on whether the teachings of Bruton applied to redactions like the ones made in this case. See infra Section IV.
In conclusion, we hold that the cutoff date for determining “clearly established Federal law” for purposes of § 2254(d)(1) is the date of the relevant state-court decision. Both the natural reading of the statutory text and post-Williams Supreme Court precedent support this conclusion. As such, Gray was not “clearly established Federal law” for the purposes of Greene’s habeas petition.
D.
Before applying our holding to the facts in this case, a brief segue is needed to address our dissenting colleague’s spirited defense of the use of the date the petitioner’s conviction became final to determine “clearly established Federal law.” While we recognize -that the issue confronted today is one over which reasonable jurists may disagree, there are some notable defi*100ciencies in the dissent’s proposed adjudication of this case. The dissent (1) would sub silentio codify Teague, including its retroactivity exceptions, as part of § 2254 without any reasoned justification for doing so, (2) erroneously asserts that Griffith v. Kentucky, 479 U.S. 314, 107 S.Ct. 708, 93 L.Ed.2d 649 (1987), applies to cases on collateral review, and (3) incorrectly asserts that our approach to § 2254(d)(1) creates a “twilight zone,” preventing a petitioner from relying on Supreme Court decisions issued after the date of his last relevant state-court decision, but before his conviction becomes final.
1.
As already explained in Section 111(A), there is direct Supreme Court precedent supporting the view that “whatever would qualify as an old rule under [the Supreme Court’s] Teague jurisprudence will constitute ‘clearly established Federal law, ... ’ under § 2254(d)(1).” Williams, 529 U.S. at 412, 120 S.Ct. 1495 (O’Connor, J., for the Court). But the dissent appears to go one step further. It alludes, at times, to the retroactive application of new rules that fall within the Teague exceptions for retroactivity as “clearly established Federal law” for purposes of § 2254.11
As a preliminary observation, this case does not raise a Teague new rules retroactivity issue. Under Teague, Gray would be an old rule since it was issued before Greene’s conviction became final. Thus, comments on the supposed benefits of Teague’s new rule retroactivity exceptions would be dicta even if we were to take the dissent’s approach. That being said, we caution that the use of Teague’s new rule retroactivity exceptions for purposes of § 2254, while not implausible, see 28 U.S.C. § 2254(e)(2)(A)© (contemplating retroactive applications of “new rule[s] of constitutional law”), id. § 2244(d)(1)(C) (same), has yet to gain support from the Supreme Court. In fact, in Horn, the Supreme Court explained that the “AED-PA and Teague inquiries are distinct.” 536 U.S. at 272, 122 S.Ct. 2147. As distinct inquiries, it is unclear whether Teague’s new rule retroactivity exceptions should be incorporated into § 2254 even if we were to adopt the use of the date the petitioner’s conviction became final for determining “clearly established Federal law.”
Indeed, the Horn decision recognized that satisfaction of § 2254(d) is the minimum required for a petitioner to receive habeas relief:
While it is of course a necessary prerequisite to federal habeas relief that a prisoner satisfy the AEDPA standard of review set forth in 28 U.S.C. § 2254(d) (“an application ... shall not be granted ... unless ” the AEDPA standard of review is satisfied (emphasis added)), none of our post-AEDPA eases have suggested that a writ of habeas corpus should automatically issue if a prisoner satisfies the AEDPA standard, or that AEDPA relieves courts from the respon*101sibility of addressing properly raised Teague arguments.
Id. Thus, under Horn, if Teague is in play at all, it is as an additional concern on top of AEDPA’s requirements codified in § 2254(d). See id. As such, it seems a leap to assume that new rules that are deemed retroactive under Teague would be automatically deemed “clearly established Federal law” for purposes of § 2254(d)(1).
2.
The dissent’s assertion that Griffith applies on collateral review cannot be reconciled with that decision’s holding. In Griffith, the Supreme Court considered whether a certain decision “applie[d] retroactively to a federal conviction then pending on direct review.” 479 U.S. at 320, 107 S.Ct. 708 (emphasis added). It held that a newly declared rule “for the conduct of criminal prosecutions is to be applied retroactively to all cases, state or federal, pending on direct review or not yet final[J” Id. at 328, 107 S.Ct. 708 (emphasis added). The principles animating Griffith were the ideas that “failure to apply a newly declared constitutional rule to criminal cases pending on direct review violates basic norms of constitutional adjudication,” id. at 322, 107 S.Ct. 708, and that courts should treat like cases alike, id. at 323, 107 S.Ct. 708.
The dissent, citing Whorton, seeks to take Griffith, a decision animated by constitutional principles pertaining to treating like cases alike on direct review, and apply it to collateral review. It sees Whorton as “explicitly” recognizing that Griffith applies to collateral review. Neither Whorton nor subsequent Supreme Court precedent support this view.
The language from Whorton upon which the dissent relies is far from explicit. See 549 U.S. at 416, 127 S.Ct. 1173. The sole citation of Griffith was for the proposition that under the “Teague framework, an old rule applies both on direct and collateral review, but a new rule is generally applicable only to cases that are still on direct review.” Id. Before assuming that the Supreme Court sought, without any additional discussion, to extend Griffith to collateral review, as the dissent suggests, a less novel understanding of the Whorton Court’s reliance on Griffith should be considered. Namely, that Griffith was probably cited as general support for the propositions that “an old rule applies ... on direct ... review [and that] a new rule is generally applicable only to cases that are still on direct review.” Whorton, 549 U.S. at 416, 127 S.Ct. 1173. The sentence following the Griffith citation in the Whorton decision further confirms this understanding by explaining how new rules apply in collateral proceedings through citation to Teague, not Griffith. Id.
Subsequent Supreme Court precedent also belies the dissent’s view that Griffith applies on collateral review. Approximately a year after Whorton, the Supreme Court, in Danforth v. Minnesota, 552 U.S. 264, 128 S.Ct. 1029, 169 L.Ed.2d 859 (2008), stated that Griffith “defined the scope of constitutional violations that would be remedied on direct appeal.” Id. at 275 n. 11, 128 S.Ct. 1029. It did so in the context of determining whether “Teague constrains the authority of state courts to give broader effect to new rules of criminal procedure than is required by that opinion.” Id. at 266, 128 S.Ct. 1029. Rather than holding that Teague applied to the state courts, like Griffith, 479 U.S. at 328, 107 S.Ct. 708, the Supreme Court reached the opposite conclusion. It held that the Teague decision did not control a state court’s decisions on retroactivity. According to the Danforth Court, the Teague decision “limits the kinds of constitutional *102violations that will entitle an individual to relief on federal habeas, but does not in any way limit the authority of a state court, when reviewing its own state criminal convictions, to provide a remedy for a violation that is deemed ‘nonretroactive’ under Teague.” Danforth, 552 U.S. at 282, 128 S.Ct. 1029. The Supreme Court emphasized that Teague, unlike Griffith, was based on the Court’s “power to interpret the federal habeas statute.” Id. at 278, 128 S.Ct. 1029. Because “Teague is based on statutory authority that extends only to federal courts applying a federal statute, it cannot be read as imposing a binding obligation on state courts.” Id. at 278-79, 128 S.Ct. 1029. While Griffith is concerned with affording individuals on direct review their right to adjudication in accord with the Constitution, Teague is derived from language in the habeas statute permitting disposal of habeas petitions “as law and justice require!.]” Id. at 278, 128 S.Ct. 1029 (citing 28 U.S.C. § 2243). Because their sources of authority are different— Griffith, the Constitution, and Teague, 28 U.S.C. § 2243 — and their motivations are different, Griffith cannot be imported wholesale into Teague without discussion. In short, we do not dispute that Griffith may somehow inform the Supreme Court’s approach in applying Teague. We also do not dispute that a § 2254 petition may invoke Griffith where a petitioner was denied the application of relevant Supreme Court precedent on direct review. But Griffith, independently, does not control retroactivity for cases on collateral review.
3.
The dissent also asserts that our approach creates a twilight zone for any petitioner who seeks to invoke Supreme Court decisions that fall between the date of the last relevant state-court decision and the date the petitioner’s conviction became final. This assertion is incorrect. Our holding does not create a categorical bar to a petitioner’s reliance on Supreme Court decisions issued during any twilight zone period. Instead, we set forth a simple rule: the universe of “clearly established Federal law” that may be applied to a particular petitioner’s § 2254 appeal is tied to the date of his last relevant state-court decision.
In this case, it was Greene’s decision not to raise the Confrontation Clause claim in his PCRA petition that established December 16, 1997, as the date of the last relevant state-court decision on the merits.12 *103This, in turn, shrank the universe of “clearly established Federal law” available to him for his § 2254 petition, relative to what that universe would have been had he pursued the Confrontation Clause claim at the PCRA stage and obtained a later, post-*104Gray state-court decision on the merits. It is unfortunate for Greene that the body of “clearly established Federal law” as of December 16, 1997, did not include the Gray decision. Yet this is an outcome he could easily have avoided by raising the Confrontation Clause claim in his PCRA petition. Doing so would have pushed the date of the last relevant state-court decision on the merits forward, thereby expanding the universe of “clearly established Federal law” to include Gray.
Using the date of the last relevant state-court decision to determine “clearly established Federal law” gives defendants incentive to pursue all colorable claims based on “Federal law” as far as possible in the state courts because doing so will give them the best chance of success in federal habeas proceedings, not to mention the underlying state proceedings. This is a salutary effect that serves Congress’s goals in passing AEDPA. See 28 U.S.C. § 2254(b) & (c) (requiring exhaustion of state court remedies); Duncan v. Walker, 533 U.S. 167, 181, 121 S.Ct. 2120, 150 L.Ed.2d 251 (2001) (referencing “clear purpose to encourage litigants to pursue claims in state court prior to seeking federal collateral review”).13
IV.
Having concluded that Gray was not “clearly established Federal law,” Greene is left to argue only Bruton and Marsh. Accordingly, we must determine whether the Pennsylvania Superior Court unreasonably applied those cases when it decided that the redactions of Finney’s and Womack’s statements did not violate Greene’s rights under the Confrontation Clause.14 In Bruton, the defendant challenged his conviction on the basis that his right to confront the witnesses against him was violated because the confession of his non-testifying codefendant, which directly implicated him, was introduced into evidence. 391 U.S. at 123, 88 S.Ct. 1620. The Court of Appeals had affirmed the defendant’s conviction because the jury had received a limiting instruction that the codefendant’s confession was competent evidence against only the codefendant. The Supreme Court reversed, holding that
because of the substantial risk that the jury, despite instruction to the contrary, looked to the incriminating extrajudicial *105statements in determining the [defendant’s] guilt, admission of [the codefendant’s] confession in this joint trial violated [the defendant’s] right of cross-examination secured by the Confrontation Clause of the Sixth Amendment.
Id. at 126, 88 S.Ct. 1620. The Court acknowledged that the instructions to the jury were clear, id. at 137, 88 S.Ct. 1620, but reasoned that
there are some contexts in which the risk that the jury will not, or cannot, follow instructions is so great, and the consequences of failure so vital to the defendant, that the practical and human limitations of the jury system cannot be ignored.....Such a context is presented here, where the powerfully incriminating extrajudicial statements of a codefendant, who stands accused side-by-side with the defendant, are deliberately spread before the jury in a joint trial.
Id. at 135-36, 88 S.Ct. 1620 (citations omitted). Accordingly, the Court refused to accept the limiting instruction as an adequate substitute for the constitutional right of cross-examination. Id. at 137, 88 S.Ct. 1620.
Two decades later, the Supreme Court revisited Bruton in Marsh. In that case, the nontestifying codefendant’s “confession was redacted to omit all reference to [Marsh]indeed, to omit all indication that anyone other than [two other individuals] participated in the crime.” 481 U.S. at 203, 107 S.Ct. 1702. Nonetheless, Marsh filed a § 2254 petition, alleging, inter alia, that the introduction of her codefendant’s confession violated her rights under the Confrontation Clause. The Supreme Court began its analysis by pointing out that Bruton was a “narrow exception” to the accepted principle that jurors are presumed to follow their instructions. Id. at 207, 107 S.Ct. 1702. It noted that there was “an important distinction between this case and Bruton, which cause[d] it to fall outside the narrow exception,” as Marsh’s codefendant’s confession did not expressly implicate her and became incriminating “only when linked with evidence introduced later at trial ([MarsbJ’s own testimony).” Id. at 208, 107 S.Ct. 1702.
The Marsh Court observed that
evidence [in a statement] requiring linkage differs from evidence incriminating on its face in the practical effects which application of the Bruton exception would produce. If limited to facially incriminating confessions, Bruton can be complied with by redactions — a possibility suggested in that opinion itself.
Id. at 208-09, 107 S.Ct. 1702. After considering the difficulties that would arise if Bruton was “extended to confessions incriminating by connection,” the Court rejected a rule that would outright prohibit the use of a codefendant’s confession. Id. at 209, 107 S.Ct. 1702. Consistent with its recognition that “confessions that do not name the defendant” are different than Bruton, which involved a facially incriminating confession, the Court held that
the Confrontation Clause is not violated by the admission of a nontestifying codefendant’s confession with a proper limiting instruction when, as here, the confession is redacted to eliminate not only the defendant’s name, but any reference to his or her existence.
Id. at 211, 107 S.Ct. 1702. In a footnote, however, the Marsh Court made clear that it was not expressing an “opinion on the admissibility of a confession in which the defendant’s name has been replaced with a symbol or neutral pronoun.” Id. n. 5.
Bruton and Marsh instruct that: (1) unredacted confessions by codefendants that implicate the defendant are generally inadmissible, Bruton, 391 U.S. at 126, 88 S.Ct. 1620; (2) redacted confessions that *106completely sanitize a codefendant’s statement by removing any reference to a defendant’s name and existence are generally admissible, Marsh, 481 U.S. at 211, 107 S.Ct. 1702; and (3) when a redaction is utilized to avoid a violation of the Confrontation Clause there must be a limiting instruction to the jury. Neither Bruton nor Marsh clarified the middle ground regarding the method for satisfactorily redacting a nontestifying codefendant’s confession. Indeed, Marsh declared that it was not addressing the admissibility of redacted statements that utilize neutral pronouns, such as the redactions of Finney’s and Womack’s statements that were admitted in Greene’s trial. Id. at 211 n. 5, 107 S.Ct. 1702.
In light of these principles, pre-Gray “courts generally followed the practice of redacting co-defendants’ statements in order to eliminate all explicit reference to other defendants on trial.” Priester, 382 F.3d at 398. Consistent with that practice, the Pennsylvania Superior Court recognized that the redaction of a codefendant’s statement was a permissible mechanism for avoiding a Bruton violation. It reviewed the redacted statements that were admitted and noted that, consistent with Bruton and Marsh, all references to the other defendants by proper name or nickname had been removed. The Pennsylvania Superior Court also noted that the jury had been properly instructed that it could only consider each redacted statement as evidence against the individual who made the statement and that these statements could not be considered in deciding any other defendant’s culpability for the crimes charged. As such, the Pennsylvania Superior Court concluded that Greene’s rights under the Confrontation Clause had not been violated.
In light of Bruton and Marsh, the Pennsylvania Superior Court’s conclusion was not unreasonable. Unlike Bruton, which involved two individuals, and Marsh, which involved three, the robbery in this case involved five or six individuals. As a result, Finney’s and Womack’s redacted statements did not directly implicate Greene. Moreover, the substitutions that were used for the names of the codefendants yielded confusing statements that failed to establish either the number of persons involved or, except for the shooter, the role that each person played in committing the offense. In the absence of a redaction that expressly implicated Greene, it was not unreasonable for the Pennsylvania Superior Court to conclude that the jury would follow the limiting instructions it had received.
V.
This case presents a vexing conundrum that cannot, no matter how one views the facts or law, be avoided. While we cannot predict with absolute certainty what date the Supreme Court would use to determine “clearly established Federal law” for purposes of § 2254(d)(1), our decision today represents a careful consideration of the pertinent, conflicting authorities, and we believe that we have reached the best conclusion given the guidance we have to date. Ultimately, only the Supreme Court can resolve such uncertainty as exists. For now, we hold that “clearly established Federal law” for purposes of § 2254(d)(1) should be determined as of the date of the relevant state-court decision. In this case, because the Pennsylvania Superior Court’s December 16, 1997 decision did not unreasonably apply the “clearly established Federal law” that existed at that time, Bruton and Marsh, we will affirm the judgment of the District Court.
. We recognize that the Pennsylvania Supreme Court granted allocatur for Greene's appeal, which raised the Confrontation Clause claim and cited Gray, but then dismissed that appeal as having been improvidently granted. That dismissal cannot be treated as the relevant state-court decision, however, because it had no precedential value:
In the circumstance where [the Pennsylvania Supreme Court has] accepted an issue by granting allowance of appeal, and [it], after conducting ... review of the issue, enters an order dismissing the appeal as having been improvidently granted, the effect is as though [it] never granted allowance of appeal. In other words, a dismissal as being improvidently granted has the exact same effect as if [it] had denied the petition for allowance of appeal (allocatur) in the first place. Where we dismiss an appeal as improvidently granted, the lower tribunal’s opinion and order stand as a decision of that court and this Court’s order has no precedential value.
Commonwealth v. Tilghman, 543 Pa. 578, 673 A.2d 898, 904 (1996) (emphasis omitted). Accordingly, even though there may have been some consideration of Greene’s arguments by the Pennsylvania Supreme Court, there was no adjudication on the merits. Thus, the Pennsylvania Superior Court decision on direct appeal is the relevant state-court decision.
. The District Court exercised jurisdiction under 28 U.S.C. §§ 2241 and 2254, and we have jurisdiction under 28 U.S.C. §§ 1291 and 2253.
. In Picard v. Connor, 404 U.S. 270, 275, 92 S.Ct. 509, 30 L.Ed.2d 438 (1971), the Supreme Court declared that "once the federal claim has been fairly presented to the state courts, the exhaustion requirement is satisfied.” It follows, therefore, that our review of whether a habeas petitioner has fairly presented his federal claim to the state courts is subject to the same standard of review that we employ in determining whether a habeas petitioner has exhausted his state court remedies. Accordingly, the question of whether Greene fairly presented his claim to the state courts is subject to plenary review. See Ellison v. Rogers, 484 F.3d 658, 660 (3d Cir.2007).
. "Since the District Court ruled on [Greene's] habeas petition without an evidentiary hearing, our review of its decision is plenary.” Thomas v. Horn, 570 F.3d 105, 113 (3d Cir.2009). We exercise plenary review over issues of statutory interpretation. United States v. Lnu, 575 F.3d 298, 300 (3d Cir.2009).
. Carey v. Musladin, 549 U.S. 70, 74, 127 S.Ct. 649, 166 L.Ed.2d 482 (2006); Yarborough v. Alvarado, 541 U.S. 652, 660-61, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004); Lockyer v. Andrade, 538 U.S. 63, 71-72, 123 S.Ct. 1166, 155 L.Ed.2d 144 (2003); Wiggins v. Smith, 539 U.S. 510, 520, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003) (O’Connor, J., joined by Rehnquist, C.J., and Stevens, Kennedy, Souter, Ginsburg, and Breyer, JJ.) (stating that AEDPA "require[s] us to limit our analysis to the law as it was 'clearly established' by [Supreme Court] precedents at the time of the state court’s decision”); but see Carey, 549 U.S. at 78, 127 S.Ct. 649 (Stevens, J., concurring) (rejecting Justice O'Connor's formulation of "clearly established Federal law”); Horn v. Banks, 536 U.S. 266, 272, 122 S.Ct. 2147, 153 L.Ed.2d 301 (2002).
. The uncertainty identified by the Supreme Court is present throughout our own admittedly contradictory § 2254(d)(1) jurisprudence. We have followed the "relevant state-court decision” approach without much fanfare in numerous cases. E.g., McMullen v. Tennis, 562 F.3d 231, 236 (3d Cir.2009); Warren v. Kyler, 422 F.3d 132, 138 (3d Cir.2005); Gibbs v. Frank, 387 F.3d 268, 272 (3d Cir. 2004). We have also used the date the petitioner’s conviction became final. E.g., Fischetti v. Johnson, 384 F.3d 140, 148 (3d Cir.2004). Indeed, it is not clear from our jurisprudence whether we recognized these divergent approaches because those cases did not require us to resolve whether the cutoff date was the relevant state-court decision date or the date the conviction became final.
. At least one of our sister circuits sees no contradiction in Justice O’Connor’s statements. In Foxworth v. St. Amand, 570 F.3d 414 (1st Cir.2009), the First Circuit concluded that Williams did not require the use of the "last reasoned state-court decision" to determine "clearly established Federal law” and endorsed the use of the date the petitioner’s conviction became final. Id. at 431, 120 S.Ct. 1495. That Court believed that Justice O'Connor expressly approved the use of Teague for determining "clearly established Federal law”:
Close perscrutation of Williams discloses nothing in the Court's constituent opinions that indicates any intention on Justice O’Connor’s part either to modify or to undercut the bright-line rule of Teague v. Lane, 489 U.S. 288, 310, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989) (effectively limiting the consideration of new constitutional rules of criminal procedure in cases on collateral review to those rules announced before the petitioner’s conviction became final).
The opposite is true. Justice O’Connor’s opinion stated that "whatever would qualify as an old rule under our Teague jurisprudence will constitute 'clearly established Federal law, as determined by the Supreme Court of the United States' under § 2254(d)(1).” 529 U.S. at 412, 120 S.Ct. 1495. That is a frank recognition that the AEDPA has neither altered nor eroded the marker laid down by Teague. This recognition is fully consistent with Part III of Justice Stevens’s lead opinion, joined by Justice O’Connor; there, Justice Stevens stated that "[t]he threshold question under AEDPA is whether [the petitioner] seeks to apply a rule of law that was clearly established at the time his state-court conviction became final." Williams, 529 U.S. at 390, 120 S.Ct. 1495 (emphasis supplied).
*96Foxworth, 570 F.3d at 431. Based on this observation, the First Circuit concluded that “finality, not the date of the last reasoned decision, is the principal determinant of whether a 'new' rule can be applied to an 'old' case.” Id. "Finality means that ‘a judgment of conviction has been rendered, the availability of appeal exhausted, and the time for a petition for certiorari [has] elapsed or a petition for certiorari [filed and] finally denied.’ ” Id. (quoting Griffith v. Kentucky, 479 U.S. 314, 321 n. 6, 107 S.Ct. 708, 93 L.Ed.2d 649 (1987)).
It cannot be denied that Justice O’Connor's approach, in light of her reference to old rules under Teague, can be read to permit the use of the date the conviction became final. That being said, the portion of the Williams decision referencing old rules under Teague has largely fallen by the wayside in post-Williams Supreme Court decisions, see infra Section III(C). Moreover, the Foxworth Court’s assertion that Justice O’Connor did not intend to "modify or to undercut the bright-line rule of Teague," 570 F.3d at 431, is not readily apparent from our reading of Williams. Justice Stevens’s claim that AEDPA codified Teague did not gamer the support of the majority. See Williams, 529 U.S. at 374-90, 120 S.Ct. 1495. Instead, the majority sided with Justice O'Connor, who rejected Justice Stevens's view that § 2254(d)(1) had "no effect on the [pre-AEDPA] law of habeas corpus[.]” Williams, 529 U.S. at 404, 120 S.Ct. 1495. Justice O’Connor's rejection of Justice Stevens’s belief that AEDPA codified Teague most certainly suggests a desire on her part to undercut Teague and the significance of the date that a petitioner’s conviction became final.
The Foxworth Court also suggested that using the date of the last relevant state-court decision would “subvert Griffith and deny criminal defendants the benefit of new Supreme Court precedent by the simple expedient of summarily affirming a lower court’s decision.” Foxworth, 570 F.3d at 432. It would also, according to the First Circuit, "give state courts a perverse incentive to avoid addressing constitutional claims in contemporaneous terms while insulating their actions from subsequent federal habeas review.” Id. We disagree. The suggestion that a state appellate court would summarily affirm the judgment of a lower state court in an effort to undermine an individual’s potential federal habeas petition is baseless. Moreover, the Teague rule itself, the rule that the First Circuit endorses, is premised on comity and respect for state court decisions, see Danforth v. Minnesota, 552 U.S. 264, 280, 128 S.Ct. 1029, 169 L.Ed.2d 859 (2008) (stating that “Teague ... was fashioned to achieve the goals of federal habeas while minimizing federal intrusion into state criminal proceedings”), not a presumption of malfeasance by state courts. To assume the latter defeats one of the main purposes of the Teague rule, to respect the finality of state-court convictions. If state courts cannot be trusted to properly adjudicate claims, then we would have no need to respect the resulting decisions.
More generally, we are unpersuaded by the First Circuit’s robust support for Justice Stevens's approach. In our view, there is no clear answer to the issue we face. According to the Foxworth Court, though, interpreting § 2254(d)(1) to require the use of the date of the relevant state-court decision is simply “untenable,” 570 F.3d at 431, as in "not able to be defended,” Memam-Webster's Collegiate Dictionary 1373 (11th ed.2003). We cannot agree with that characterization. See infra Sections III(B) and (C). If Griffith truly "remove[d] any vestige of doubt,” 570 F.3d at 431 (emphasis added), and Teague and Griffith together "le[d] to the inexorable conclusion,” id. at 432 (emphasis added), that the date the petitioner’s conviction became final is the correct date, why then, has the Supreme Court expressed uncertainty in its recent decisions? See Spisak, 130 S.Ct. at 681; see also Thaler, 130 S.Ct. at 1174 n. 2. Perhaps the resolution of this issue by the Supreme Court would be a simple task, but resolution of the issue in the Courts of Appeals based on existing Supreme Court precedent is akin to trying to piece together a jigsaw puzzle that has been sprinkled with pieces from other puzzles. All the pieces, no matter how they are arranged, simply do not fit.
. Justice Stevens's reliance on the date the petitioner’s conviction became final appears to be based on his position that § 2254(d)(1) codified Teague. See Williams, 529 U.S. at 380, 120 S.Ct. 1495.
. At a minimum, when the state properly raises Teague, the federal court must conduct the Teague analysis. Horn, 536 U.S. at 272, 122 S.Ct. 2147. Notably, the Commonwealth did not raise Teague in this case.
. Notably, the phrase "clearly established Federal law” did not have any special meaning for federal habeas review prior to AEDPA and it bears no overt connection to any language used in Teague.
. "In Teague, [the Supreme Court] explained that unless 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.” Horn, 536 U.S. at 271, 122 S.Ct. 2147 (internal quotation marks omitted). There are two exceptions to the general rule. "The first exception permits the retroactive application of a new rule if the rule places a class of private conduct beyond the power of the State to proscribe ... or addresses a substantive categorical guarantee accorded by the Constitution[.]” Id. n. 5 (internal quotation marks omitted). "The second exception is for watershed rules of criminal procedure implicating the fundamental fairness and accuracy of the criminal proceeding.” Id. (internal quotation marks omitted).
. The dissent asserts that Greene would not have been able to raise his Confrontation Clause claim in a PCRA petition because the issue had been previously litigated on the merits by the Pennsylvania Superior Court. See Pa. Cons.Stat. § 9543(a)(3); id. § 9544(a). We disagree. ''[T]he 'previously litigated' rule codified in § 9544(a) simply relieves Pennsylvania courts of the burden of revisiting issues which are res judicata.” Boyd v. Warden, 579 F.3d 330, 369 (3d Cir.2009) (en banc). The Pennsylvania Superior Court’s decision on direct review held only that Greene’s "right to confrontation as set forth in Bruton " was not violated. The Confrontation Clause claim presented in Greene’s PCRA petition would have been based on Gray, a "discrete legal ground,” Commonwealth v. Collins, 585 Pa. 45, 888 A.2d 564, 570 (2005), for relief that was separate from the grounds set forth in Bruton and Marsh. See Marsh, 481 U.S. at 211 n. 5, 107 S.Ct. 1702(stating that the Supreme Court "express[ed] no opinion on the admissibility of a confession in which the defendant's name has been replaced with a symbol or neutral pronoun”).
Unlike the comparatively loose fair presentation requirement for a § 2254 petition, which requires that the habeas petitioner need merely "present [the] federal claim's factual and legal substance to the state courts in a manner that puts them on notice that a federal claim is being asserted,” McCandless, 172 F.3d at 261, the PCRA requirement that an issue not have been previously litigated “refers to the discrete legal ground raised and decided on direct review,” Collins, 888 A.2d *103at 570. While raising Bruton on direct appeal to the Pennsylvania Superior Court was sufficient to put the state courts on notice that a federal claim was being asserted, satisfying the fair presentation requirement, McCandless, 172 F.3d at 261, the Pennsylvania Superior Court's decision was limited to whether Bruton required that Greene have been granted a severance at trial. Thus, for purposes of the PCRA, that decision did not bear on whether Gray, "a discrete legal ground” that was not "decided on direct review,” Collins, 888 A.2d at 570, would have provided Greene relief.
In addition, we do not consider the authorities cited by the dissent persuasive given the facts of this case. The dissent cites Commonwealth v. Small, 602 Pa. 425, 980 A.2d 549 (2009), for the proposition that reliance on a "different case to support the original theory,” id. at 569, of relief would not "change the fact that [an] issue was previously litigated,” id. But that proposition was offered based on different factual circumstances. In Small, the petitioner was seeking to rely on United States Supreme Court authority that existed at the time of his direct appeal yet was not raised in that appeal to effectively re-argue an issue that was previously litigated on direct appeal. Id. The Small Court had no reason to ,view the issue as novel because the underlying United States Supreme Court authority existed at the time of the direct appeal. Here, Gray did not exist at the time of the Pennsylvania Superior Court’s decision. As such, Greene would not be simply raising a theory or allegation in support of a discrete legal ground for relief that existed at the time of his direct appeal. Instead, he would be asserting a new discrete legal ground for relief based on previously unavailable United States Supreme Court precedent.
The dissent’s reliance on Commonwealth v. Washington, 592 Pa. 698, 927 A.2d 586 (2007), is similarly misplaced. There, the petitioner argued that Gray, which had been decided after his direct appeal to the Pennsylvania Supreme Court but before that court’s decision became final, entitled him to relief under the PCRA. Id. at 608-09. The Pennsylvania Supreme Court found that the petitioner's Confrontation Clause claim had been previously litigated on the merits, but its conclusion was not based on the fact that it had already considered the petitioner’s Bruton claim. Id. at 609. On direct review, the Court assumed arguendo that the petitioner's right to confrontation had been violated by the use of the word “blank” in a redacted statement and then held that the error was harmless and did not prejudice him. Id. In other words, it assumed that the rule later set forth in Gray existed and held that, despite the violation of that rule, the petitioner could not receive relief. In its PCRA review, the Court took note of Gray, which supported the proposition of law it had already assumed on direct appeal, and concluded that it had already considered the substance of that claim on direct review. Id. As such, Washington does not support the dissent’s view that Gray and Bruton raise the same issue for purposes of PCRA review.
Finally, we note that the PCRA’s requirement that an issue not have been previously litigated is flexibly applied by the Pennsylvania courts to avoid injustice, not create it. The Pennsylvania legislature, "by its effort to channel the broadest category of post-conviction claims into the statutorily-prescribed procedures ... implemented a scheme that must necessarily be deemed to take into account facets of traditional habeas corpus jurisprudence ... under which previous litigation does not function as a never-yielding bar to the possibility of collateral relief.” Commonwealth v. Cruz, 578 Pa. 263, 851 A.2d 870, 877 (2004) (citations omitted). ”[I]n ... instances involving unique circumstances embodying manifest error or irregularity in the chain of previous litigation, th[e] [Pennsylvania Supreme] Court and others have found that the doctrine need not be regarded as dispositive on collateral review.” Id. Thus, even if we credit the dissent's view that Greene’s issue was previously litigated, his PCRA petition was not doomed from its outset. The Pennsylvania Supreme Court's dismissal of Greene’s petition as improvidently granted on direct review, in combination with the United States Supreme Court’s issuance of the Gray decision, could be the kind of "irregularity in the chain of previous litigation,” id., that would merit relaxation of the requirement that an issue not have been previously litigated.
. Notably, our approach does not create a procedural bar to Greene's claim. As explained in Section II, Greene satisfied the exhaustion requirement by fairly presenting his Confrontation Clause claim to the Pennsylvania Superior Court. Satisfaction of the exhaustion requirement, however, does not open the door to Supreme Court decisions issued after the last relevant state-court decision. The determination of what constitutes "clearly established Federal law” for purposes of § 2254 is a merits-based determination. Whether a petitioner has fairly presented and exhausted his claim in the state courts, in contrast, is a procedural issue that we decide independent of the merits of the petitioner’s claim.
. "[T]he 'unreasonable application' prong of 28 U.S.C. § 2254(d)(1) requires a habeas petitioner to show that the state court's application of Supreme Court precedent was 'objectively unreasonable.' ” Fountain v. Kyler, 420 F.3d 267, 273 (3d Cir.2005). “[A] federal court's mere disagreement with, the state court's application of Supreme Court precedent is not sufficient; rather, a state court adjudication fails the 'unreasonable application’ test only if the state court identified the correct governing legal rule but unreasonably applied it to the particular case or if the state court either unreasonably extended a legal principle from Supreme Court precedent to a new context in which it should not apply or where it unreasonably refused to extend such a principle to a new context in which it should apply.” Id. We do not address the "contrary to” prong of § 2254(d)(1) as the Pennsylvania Superior Court correctly identified Bruton as the controlling legal authority. See Williams, 529 U.S. at 406, 120 S.Ct. 1495.