dissenting:
To prevail on a claim under Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), a habeas petitioner must show that his attorney performed deficiently and that such performance prejudiced him. This case presents an important, recurring issue under Strickland’s prejudice prong: the proper standard for evaluating whether a petitioner was prejudiced by his attorney’s failure to present mitigating evidence at sentencing. When evaluating a failure-to-present-mitigating-evidence claim under Strickland’s prejudice prong, Supreme Court law is clear that a court must consider not just the benefits of never-presented mitigating evidence, but also its drawbacks. *1177The district court did not apply that prejudice standard here. Yet, rather than remand this case to let the district court apply the right standard, the panel majority tries to apply that standard for the first time on appeal.
With respect, that effort has my colleagues flailing in the dark. The record before us is huge and the prejudice issue in this case is especially difficult. We appellate judges are ill-suited to apply the correct prejudice standard, in the first instance, to a voluminous record. A district court decision applying the right standard would have been invaluable to our review of this challenging case.
Rather than evaluating prejudice on the merits, I would clarify the standard for evaluating failure-to-present-mitigating-evidence claims under Strickland, then remand this case, once again, to allow the district court to reevaluate prejudice under the correct standard. This would do a service to circuit law — which has long been confused on that standard — and would aid us greatly if we were to consider this case on another occasion. Thus, I respectfully must dissent from the majority’s decision to take a different approach.
I
■When we first considered this case, we said that “[a] more complete presentation” of evidence at the penalty phase “could have made a difference” in Stankewitz’s sentence. Stankewitz v. Woodford, 365 F.3d 706, 724 (9th Cir.2004). Applying that instruction on remand, the district court ruled that, “[sjince many of Stankewitz’s [mitigation] allegations are proved by official documents in the record, the requirements for his ineffective assistance of counsel claim as set forth in [our 2004] opinion are satisfied.” In so ruling that Stankewitz had suffered prejudice from his attorney’s performance, however, the district court failed to evaluate the potentially aggravating effect of much of Stankewitz’s new evidence.
The district court therefore did not apply the correct legal standard. As the Supreme Court recently reaffirmed in Wong v. Belmontes, when evaluating Strickland prejudice “it is necessary to consider all the relevant evidence that the jury would have had.” 558 U.S. 15, 130 S.Ct. 383, 386, 175 L.Ed.2d 328 (2009) (per curiam). A court may not consider just the mitigating evidence — or just the mitigating effect of any evidence — that might have been presented. See id. at 386, 390. Ostensibly mitigating evidence, after all, can be a “double-edged sword”: it can also be aggravating or can invite devastating rebuttal. See, e.g., Cullen v. Pinholster, — U.S. -, 131 S.Ct. 1388, 1410, 179 L.Ed.2d 557 (2011) (evidence regarding petitioner’s family problems may not have been mitigating because it could have led the jury to conclude that the petitioner was beyond rehabilitation); Atkins v. Virginia, 536 U.S. 304, 321, 122 S.Ct. 2242, 153 L.Ed.2d 335 (2002). While mitigating evidence might “ma[k]e a difference” if presented, it can do so “in the wrong direction” for the habeas petitioner. Belmontes, 130 S.Ct. at 388.
That may well be the case here. Stankewitz has offered evidence of his difficult youth, of his history of mental illness, and of his substance abuse and lack of sleep before the murder. But much of this evidence “is of questionable mitigating value,” Pinholster, 131 S.Ct. at 1410, and — worse yet — could well have worked “in the wrong direction” for Stankewitz, Belmontes, 130 S.Ct. at 388. As a youth Stankewitz was a violent sociopath who often hurt people. Evidence of his youth was thus “by no means clearly mitigating, as the jury might have concluded that [Stankewitz] was simply beyond rehabilitation.” Pinholster, 131 S.Ct. at 1410. The same goes for the *1178evidence of mental illness and of drug abuse, which likewise may have done Stankewitz more harm than good. See id.; Brewer v. Quarterman, 550 U.S. 286, 289-90, 292-93, 127 S.Ct. 1706, 167 L.Ed.2d 622 (2007) (evidence of mental illness and substance abuse can be a double-edged sword); Correll v. Ryan, 539 F.3d 938, 963 (9th Cir.2008) (O’Scannlain, J., dissenting) (collecting cases explaining that evidence of mental illness may harm more than help).
Yet the district court weighed the mitigating evidence wholly in Stankewitz’s favor. In doing so, the court apparently believed that it was following our direction that “even a fraction of the details Stankewitz [previously] allege[d] could have made a difference” in his sentence. Stankewitz, 365 F.3d at 724. But whatever we last said about the potential effect of Stankewitz’s allegations, the district court was still required to weigh both the good — and the bad — effects of Stankewitz’s new evidence. That is what Strickland requires and what Wong v. Belmontes reaffirms.
II
Because the district court did not apply the correct legal standard, on appeal we have been deprived of a solidly grounded district court decision to aid us. Yet this is no matter for the majority, which, for the first time on appeal, purports to apply the right standard. But the majority’s thin prejudice analysis shows only how hard it is to evaluate Strickland prejudice when the district court does not, in the first instance, apply the correct standard to an enormous factual record.
A
The majority declines to conclude “that Stankewitz was not prejudiced by [his counsel’s] failure to investigate or present any of the mitigating evidence” largely because “Stankewitz’s case is materially different from [Wong v..] Belmontes.” Maj. op. 1174. This case is not as bad as Belmontes, the majority contends, because (1) the aggravating effect of Stankewitz’s mitigation evidence might not have been as bad as in Belmontes, and (2) in contrast to Belmontes, Stankewitz’s jury had already gotten a taste of that aggravating effect. Maj. op. 1173-74.
These distinctions may be accurate, but they do not show that Stankewitz was prejudiced. The majority’s analysis shows only that this case may be closer than Belmontes. That does not say much. The decision in Belmontes was both unanimous and summary. It was not close. Belmontes should not be treated as though it marked the exact boundary between prejudice and harmlessness. This case may present a closer call, but that does not mean that the call must be made for Stankewitz.
B
Once we step past the majority’s efforts to distinguish Wong v. Belmontes, we are left with almost no analysis of prejudice.1 The majority lobs some conclusory assertions that Stankewitz suffered prejudice. See, e.g., maj. op. 1170-71 (contending, without analysis, that evidence of Stankewitz’s “life history” would have tipped the *1179scale). But the majority gives no persuasive grounds for concluding that Stankewitz has demonstrated a reasonable probability that his sentence would have been different with his mitigation evidence.
Rather than peddling an unpersuasive prejudice analysis, we should have used this case to clarify the law on double-edged mitigation evidence and then remanded to allow the district court to evaluate prejudice under the right standard. That approach is common to federal appellate combs. See, e.g., United States v. Bus. of Custer Battlefield Museum & Store Located at Interstate 90, Exit 514, S. of Billings, Mont., 658 F.3d 1188, 1196 (9th Cir.2011) (Fisher, J.) (vacating and remanding for the district court to apply the correct legal standard); Tamas v. Dep’t of Soc. & Health Sens., 630 F.3d 833, 837, 847 (9th Cir.2010) (same); cf. Sears v. Upton, — U.S. -, 130 S.Ct. 3259, 3261, 3265-67, 177 L.Ed.2d 1025 (2010) (per curiam) (vacating and remanding when lower court had not applied the correct prejudice inquiry); United States v. Lanier, 520 U.S. 259, 272, 117 S.Ct. 1219, 137 L.Ed.2d 432 (1997) (“Because the Court of Appeals used the wrong gauge in [reaching its decision] ..., we vacate the judgment and remand the case for application of the proper standard.”).
This approach is especially warranted where, as here, a case demands a difficult weighing of voluminous evidence — an exercise at which district courts excel. Rather than reach out to decide the merits, I would simply remand to let that exercise take its course. That is the prudent approach.
Ill
That approach also would have done a service to our circuit law.
Before the decision in Wong v. Belmontes, our circuit was awash with Strickland cases that ignored the drawbacks of ostensibly mitigating evidence. E.g., Corred v. Ryan, 539 F.3d 938 (9th Cir.2008); Belmontes v. Ayers, 529 F.3d 834 (9th Cir.2008); Ainsworth v. Woodford, 268 F.3d 868 (9th Cir.2001). This reasoning squarely conflicted with the Supreme Court’s Strickland jurisprudence. See, e.g., Corred, 539 F.3d at 962 (O’Scannlain, J., dissenting) (explaining that, in finding Strickland prejudice, the panel majority “ignores a mountain of precedent which requires us to consider not only the benefits of the ostensibly mitigating evidence counsel failed to present, but also its potential drawbacks”); Belmontes v. Ayers, 529 F.3d at 879-80 (O’Scannlain, J., dissenting) (“In order to discern prejudice, the majority overstates the mitigating evidence, understates the properly admitted aggravating evidence, and ignores the further aggravating evidence that would have come in on rebuttal.”); Ainsworth, 268 F.3d at 880 (Graber, J., dissenting) (“[M]uch of the mitigating evidence on which the majority relies presented a double-edged sword, opening the door to harmful rather than helpful inferences.”).
The Supreme Court took notice and repudiated this reasoning in Wong v. Belmontes. Yet our circuit is already showing signs of backsliding. See, e.g., James v. Ryan, 679 F.3d 780, 810-20 (9th Cir.2012) (finding Strickland prejudice based on evidence of the petitioner’s dysfunctional upbringing, drug abuse, and mental illness — without citing Wong v. Belmontes or considering the potential downside of such evidence). We should have halted that backsliding by making clear that Wong v. Belmontes invalidates significant parts of our circuit law regarding failure-to-present-mitigating-evidence claims under Strickland.
W
While I hesitate to remand this decades-old case once more, that is the prudent *1180course. In this capital case, the stakes are high enough to take the long view, get circuit law right, and leave Stankewitz’s particular case for another day.2 If that day were to come, we would have the benefit of a better-grounded district court decision, and we could be confident that we got this case right. We cannot be confident today.
I respectfully dissent.
. The majority does attempt briefly to analogize this case to Wiggins v. Smith, 539 U.S. 510, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003), see maj. op. 1174-75, but Wiggins is distinguishable for the reason the majority gives: “because Wiggins did not have a pattern of aggressive behavior or a criminal history, his mitigation evidence was unlikely to pose the double-edged sword problem presented in other cases (and now exemplified by Belmontes)." Maj. op. at 1175 (citing Wiggins, 539 U.S. at 515-16, 535-36, 123 S.Ct. 2527); see also id. (admitting that “Stankewitz's history is certainly not benign like Wiggins' ").
. It would also have relieved us from delving into whether Stankewitz’s trial counsel performed deficiently. The Supreme Court has stated that, ”[i]f it is easier to dispose of an ineffectiveness claim on the ground of lack of sufficient prejudice, ... that course should be followed.” Strickland, 466 U.S. at 697, 104 S.Ct. 2052; see Belmontes, 130 S.Ct. at 386 (a court need not resolve whether counsel performed deficiently when a petitioner cannot establish prejudice). Rather than muscling through the record to evaluate performance, the majority should have waited to see if it needed to evaluate performance at all. Because I believe that we should have waited, I offer no view on the majority’s analysis of Strickland's performance prong.