Opinion by Chief Judge KOZINSKI; Dissent by Judge REINHARDT.
KOZINSKI, Chief Judge:On remand from the Supreme Court, Uttecht v. Brown, — U.S.-, 127 S.Ct. 2218, 167 L.Ed.2d 1014 (2007), we consider whether defense counsel’s performance was deficient.
Facts
After raping and murdering Holly Washa,1 Brown was convicted of aggravated first-degree murder in Washington. With the aid of an investigator, a social worker and a mitigation specialist, Brown’s three experienced attorneys put on a thorough mitigation case during the penalty phase of Brown’s trial. They (1) created a 250-page life chronology detailing Brown’s social and medical history, and presented most of this information at trial; (2) introduced evidence that Brown had a mental disorder; (3) called Dr. Maiuro, a clinical psychologist; and (4) called multiple character witnesses, such as family members. Nonetheless, the jury sentenced Brown to death.
After exhausting his direct appeals and state collateral review, Brown petitioned for a writ of habeas corpus in federal court, raising a number of constitutional claims regarding his trial and sentencing.2 *1033The district court denied his petition after holding an evidentiary hearing. Brown appeals on three issues relating to his death sentence: the facial validity of the Washington death penalty statute, the exclusion of jurors and ineffective assistance of counsel. We also expanded the certificate of appealability to include whether the district court erred in excluding death penalty trial reports.
We upheld Washington’s death penalty statute, Brown v. Lambert, 451 F.3d 946, 947-48 (9th Cir.2006), but ruled that a juror was unconstitutionally excluded, id. at 948-54. The Supreme Court then reversed us on the juror exclusion issue. Uttecht, 127 S.Ct. at 2222. We therefore affirm the district court’s rulings that the Washington death penalty statute is facially valid, see Brown v. Lambert, 451 F.3d at 947-48, and that the jury selection for Brown’s trial was constitutional, see Uttecht, 127 S.Ct. at 2222. We now address Brown’s ineffective assistance of counsel claim, including the district court’s exclusion of the death penalty trial reports.
Analysis
1. To establish ineffective assistance of counsel, Brown must show that defense counsel’s performance was objectively deficient and prejudiced his defense. See Edwards v. Lamarque, 475 F.3d 1121, 1126 (9th Cir.2007) (en banc) (citing Strickland v. Washington, 466 U.S. 668, 687, 694, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984)). Under AEDPA, Brown must also show that the state court adjudication was objectively unreasonable. Id. at 1125-26 (citing 28 U.S.C. § 2254(d)). Brown’s lawyers presented significant mitigating evidence, unlike other cases where counsel were deficient for presenting hardly any mitigation case at all. Cf., e.g., Rompilla v. Beard, 545 U.S. 374, 381-86, 125 S.Ct. 2456, 162 L.Ed.2d 360 (2005); Frierson v. Woodford, 463 F.3d 982, 989-93 (9th Cir.2006). Nevertheless, Brown argues that his representation was unconstitutionally deficient because his lawyers did not (1) call a psychiatrist, (2) call Sally Schick, Brown’s former prison counselor, and (3) cross-examine the prosecution’s psychiatrist, Dr. Brinkley.
a. Failure to call a psychiatrist
More than nine months before trial, defense counsel learned that Brown may have had a mental disorder. Acting on the advice of a neuropsychologist, counsel retained Dr. Maiuro, a well-respected clinical psychologist who diagnosed Brown as manic, and as suffering from antisocial personality disorder and sexual sadism. Counsel designed an extensive mitigation case based on the theory that Brown’s troubled home life had led to these mental disorders, and that had Brown been properly treated upon his release from prison shortly before the killing, he might not have committed the crime. At the penalty phase trial, counsel presented witnesses to testify at length regarding Brown’s upbringing and social history, and also put on extensive testimony from Dr. Maiuro concerning Brown’s mental problems. Counsel debated whether to also retain a psychiatrist, but they ultimately concluded that a psychiatric evaluation wouldn’t provide as much useful information as a psychological evaluation, including the administration of formal psychological testing. At trial, Dr. Maiuro competently testified that Brown suffered from all three mental disorders and explained why he had diagnosed them. He further testified that manic disorder was generally treatable with lithium.
*1034Dr. Maiuro was qualified to diagnose Brown’s mental condition and testify that Brown could have been treated with lithium. But Dr. Maiuro couldn’t prescribe lithium as he isn’t a physician. Consequently, just before trial, Dr. Maiuro recommended that defense counsel consult a psychiatrist, and specifically recommended Dr. Brinkley. But when defense counsel approached Dr. Brinkley, they learned that he had already agreed to testify for the prosecution. Because it was so close to trial and defense counsel had previously decided not to retain a psychiatrist, they didn’t contact any other psychiatrists. At trial, Dr. Brinkley testified that, after his review of Brown’s medical records, he could see no basis for prescribing lithium, as there was “no clear indication” that Brown had “a disorder for which lithium was appropriate.”
The prosecution’s closing argument relied on Dr. Brinkley’s testimony to argue that Dr. Maiuro was “kind of out on a limb” in the part of his testimony concerning lithium. The prosecution did not, however, argue that Dr. Maiuro was wrong in his diagnosis of manic disorder. It only reminded the jury that “in Dr. Brinkley’s opinion,” Brown “did not have any sort of disorder that ... would suggest that lithium would be appropriate.” It also emphasized Dr. Maiuro’s testimony that Brown was a sexual sadist who “took pleasure ... in some way [from] sex and violence,” and did not question Dr. Maiuro’s diagnosis of antisocial personality disorder. Defense counsel responded that if Brown should not have been on lithium, then the fact that Oregon had him take lithium could have caused “vast and far reaching complications” affecting his mental state, thus reducing his culpability. Defense counsel also emphasized Brown’s “sexual sadism” as a mitigating factor because it arose from his difficult childhood. (Brown does not argue that his counsel should not have introduced the sexual sadism evidence.)
Brown argues that counsel were deficient for failing to call a psychiatrist to rebut Dr. Brinkley’s 'testimony that lithium wouldn’t have helped Brown. Brown suggests that only someone who was able to prescribe lithium himself could have affirmatively rebutted Dr. Brinkley’s testimony that Brown did not have a condition that was treatable with lithium. But Dr. Maiuro was qualified to testify about lithium, even though he was not himself licensed to prescribe it, and everyone agrees that he was widely respected. The additional weight, if any, of testimony by a psychiatrist was outweighed by other considerations that defense counsel took into account. Retaining a psychiatrist would have required a continuance. Counsel offered two legitimate reasons for not seeking a continuance: They wanted the jury to deliberate over the Christmas holiday, when jurors might be more merciful, and they wanted to give the prosecution less time to prepare its penalty phase case.
In any event, it’s far from clear that Dr. Brinkley’s testimony was particularly damaging to defense counsel’s mitigation theory. While Dr. Brinkley cast some doubt on whether Brown had a mental disorder that was treatable with lithium, he did not ultimately dispute the larger defense theory — that Brown developed serious mental problems as a result of his difficult family history. As defense counsel put it at closing argument, a pattern of ongoing abuse led to “adult disorders [that] created major problems” for Brown. Defense counsel’s theory was supported by an investigation yielding hundreds of pages of material, and was bolstered at trial through extensive testimony from Brown’s family members. This foundation was not seriously undermined by Dr. Brinkley.
Nor, as we have noted, did Dr. Brinkley squarely dispute Dr. Maiuro’s diagnosis of manic disorder, or cast any doubt on the *1035diagnoses of antisocial personality disorder and sexual sadism. Defense psychiatric testimony might or might not have convincingly rebutted Dr. Brinkley’s views on the usefulness of lithium for Brown’s mental state. But it would not have greatly changed the mitigation case put before the jury.
Furthermore, retaining a psychiatrist involved significant risk. The trial court had ruled that, after the guilt phase verdict, defense counsel were required to give the prosecution all written reports they had concerning Brown’s mental health. See State v. Pawlyk, 115 Wash.2d 457, 800 P.2d 338, 349-50 (1990) (en banc). By retaining a psychiatrist, therefore, defense counsel would have risked obtaining unfavorable written reports (like Dr. Brinkley’s), which they would then have had to turn over to the prosecution. This was not a trivial risk, as several of the mental health experts who evaluated Brown had not diagnosed him with manic disorder— the theory advanced by the defense.
Using the “ ‘fabled twenty-twenty vision of hindsight,’ ” Edwards, 475 F.3d at 1127 (quoting LaGrand v. Stewart, 133 F.3d 1253, 1271 (9th Cir.1998)), we now know that Brown’s habeas counsel eventually found Dr. Scher, a psychiatrist who could testify that Brown had bipolar disorder and could have been treated with lithium. But Brown’s trial counsel couldn’t have known Dr. Scher’s ultimate opinion, as she didn’t reach her conclusion until almost a decade after the trial.3
Ultimately, it doesn’t matter whether we agree with trial counsel’s decision not to obtain a psychiatric evaluation after Dr. Maiuro recommended that they do so. What matters is that defense counsel used “sound trial strategy,” Edwards, 475 F.3d at 1126 (quoting Strickland, 466 U.S. at 689, 104 S.Ct. 2052), in responding to Dr. Maiuro’s suggestion. Counsel did not ignore the suggestion; they followed up on it and even contacted Dr. Brinkley, the psychiatrist that Dr. Maiuro recommended as likely to give a favorable evaluation. When Dr. Brinkley proved unavailable and agreed to be a witness for the state, the difficulties and risks of obtaining other psychiatric evaluations became even more acute. We can’t say that trial counsel weighed them in an irrational or unprofessional manner — or even that they made the wrong choice. The district court did not err in finding that petitioner has not shown ineffective assistance of counsel on this point.
b. Failure to call Sally Schick
Schick is a licensed professional counselor who evaluated Brown regularly for two years while she was working at an Oregon prison where Brown had previously been incarcerated. She believed that Brown had a mental disorder, and requested that Brown be treated with lithium. Although at least one Oregon prison system psychiatrist disagreed with Schick’s diagnosis, she found one who prescribed him lithium for a five-to-six-month-long tri*1036al period. Schick only observed Brown for a month after the trial started, so she didn’t know whether Brown completed the lithium trial. During the time she watched Brown, she saw at most a “little bit” of “gradual change” in his behavior. After his release, Brown chose to stop taking lithium before the amount he was given on release ran out. -
Counsel contacted Schick before trial, and decided not to call her as a witness. Instead, they introduced Schick’s notes and Dr. Maiuro discussed them during his testimony. Counsel were surely not incompetent in making this use of Schick’s evaluation of Brown, rather than having her take the stand. Schick was not an M.D. or a Ph.D.; she was not like Dr. Maiuro — a university professor and a widely-published author in his field, who was viewed even by Dr. Brinkley as highly competent. She was merely a licensed professional counselor, who was not competent even to perform psychological testing. Indeed, an Oregon psychiatrist had explicitly disagreed with her recommendation that Brown should take lithium. Furthermore, had Schick testified, her testimony would have been challenged on the same basis as Dr. Maiuro’s, as she could not prescribe lithium. Worse yet, Schick could have been impeached with treatment reports she filled out after observing the first month of Brown’s trial, in which she indicated that despite the lithium trial, there had been, at worst, “no change” in his condition — and only “slight improvement” at best. Plus, the prosecution could have forced her to make various damaging concessions, including that she had done nothing to determine whether Brown was malingering and that, even if her diagnosis was correct, manic disorder does not “cause somebody to commit sexually violent crimes.”
Counsel did not ignore or overlook the possibility of calling Schick to the stand and made a reasonable judgment to use her notes instead. We can’t say that this was unconstitutionally deficient representation.
c. Cross-examination of Dr. Brinkley
Brown argues that cross-examining Dr. Brinkley could have established that he was biased because he didn’t interview Brown before making his diagnosis, and forced Dr. Brinkley to concede that Dr. Maiuro was competent to render a psychological diagnosis.
We give “great deference” to “counsel’s decisions at trial, such as refraining from cross-examining a particular witness.” See Dows v. Wood, 211 F.3d 480, 487 (9th Cir.2000). Applying that standard, we cannot say that defense counsel’s decision was objectively unreasonable. Defense counsel were prepared to cross-examine Dr. Brinkley, but they made the tactical decision not to. Dr. Brinkley’s conclusions were based solely on records from the Oregon prison, and defense counsel’s theory was that Oregon officials had not properly treated Brown’s mental disorder. Counsel therefore reasonably believed that Dr. Brinkley’s testimony didn’t harm their defense. By not cross-examining Dr. Brinkley, they avoided the risk that his testimony would conflict with their own evidence, which showed that Brown had manic disorder. Dr. Brinkley testified at the evidentiary hearing that he believed Brown was not manic but was, instead, a sociopath. The jury may not have bought defense counsel’s theory, but counsel certainly weren’t unreasonable for advancing it, or for avoiding any direct contradiction of it by not cross-examining Dr. Brinkley.
The evidentiary record developed on habeas supports defense counsel’s determination at trial that a cross-examination of Dr. Brinkley “might well have backfired.” Yarborough v. Gentry, 540 U.S. 1, 7, 124 *1037S.Ct. 1, 157 L.Ed.2d 1 (2003) (per curiam). Indeed, during oral argument before us, Brown’s habeas counsel conceded that the cross-examination at Dr. Brinkley’s deposition (which was taken for the district court’s evidentiary hearing) could have hurt Brown’s mitigation case had it been conducted at trial. See Oral Argument at 6:13. After all, the record indicates that the prosecution did not have Dr. Brinkley interview Brown because defense counsel was challenging his authority to do so. And, in any event, Dr. Brinkley testified at the evidentiary hearing that he had no interest in doing so, because he thought Brown was such a liar that any interview would be pointless. Dr. Brinkley also testified that the Oregon prison records were sufficient to diagnose Brown. Furthermore, at the evidentiary hearing, Dr. Brinkley made his view clear that, even if Brown were manic, he was not suffering from a manic episode at the time he raped and murdered Washa. Dr. Brinkley also testified that manic episodes are not associated with the sort of premeditated behavior Brown displayed. So even if Brown had been manic and on lithium, the lithium would not, in his view, have prevented the crime. It was reasonably foreseeable that Dr. Brinkley would have come up with similarly damaging testimony if cross-examined at trial. Brown’s counsel therefore weren’t deficient for failing to cross-examine Dr. Brinkley. Rather, they made a tactical decision, based on their theory of the case, not to give Dr. Brinkley an opportunity to undermine that theory.
2. The district court didn’t abuse its discretion, Tritchler v. County of Lake, 358 F.3d 1150, 1155 (9th Cir.2004), in excluding the death penalty trial reports, which were essentially summaries of other Washington death penalty cases. Prior to the district court’s evidentiary hearing, the parties were required to submit a pre-trial order identifying all exhibits. Brown didn’t identify the reports in the pre-trial order, and he didn’t offer them during the evidentiary hearing. Instead, he presented the reports as attachments to his written closing statement. The prosecution moved to strike the reports and, not surprisingly, the district court granted the motion.
Even putting aside a district court’s broad “discretion to exclude an exhibit not identified in the pretrial order,” Swinton v. Potomac Corp., 270 F.3d 794, 809 (9th Cir.2001), evidence must surely be proffered by the time of the evidentiary hearing, so that the other side can have a fair opportunity to address or rebut it. The district court certainly did not abuse its discretion in failing to consider evidence that was so untimely.
* * *
Brown’s counsel weren’t objectively deficient, as they made reasonable strategic decisions by not calling a psychiatrist, not calling Sally Schick and not cross-examining Dr. Brinkley. The district court therefore correctly rejected Brown’s ineffective assistance of counsel claim. Nor did the district court abuse its discretion in excluding the death penalty trial reports.
AFFIRMED.
. For a more detailed discussion of the facts, see the Washington Supreme Court’s opinion in Brown’s direct appeal, State v. Brown, 132 Wash.2d 529, 940 P.2d 546, 555-59 (1997) (en banc).
. Because Brown filed his habeas petition after April 23, 1996, we apply the "substantive review standards of the Anti-terrorism and Effective Death Penalty Act of 1996, Pub.L. No. 104-132, 110 Stat. 1214 (1996) ('AED-PA’).” Webster v. Woodford, 369 F.3d 1062, *10331066 (9th Cir.), cert. denied, 543 U.S. 1007, 125 S.Ct. 626, 160 L.Ed.2d 471 (2004); see also Lindh v. Murphy, 521 U.S. 320, 327, 117 S.Ct. 2059, 138 L.Ed.2d 481 (1997).
. Even if Dr. Scher had testified, the prosecution would have forced her to make many damaging concessions during cross-examination. Dr. Scher admitted that Brown knew right from wrong, and that he was in control of his behavior. She is not a forensic psychiatrist, and has no expertise in sexual sadism. In fact, she was aware of no literature supporting her view that lithium treatment would have made Brown less likely to commit a sexually violent crime, nor did she know of any tests or studies that she could have done to reinforce that opinion. Furthermore, the prosecution’s argument that Dr. Brinkley was best qualified to discuss connections between lithium and Brown’s crime could have been strengthened had Dr. Scher testified, because she had consulted with Dr. Brinkley about proper medications in other circumstances. It’s therefore not clear that Dr. Scher’s testimony would have helped Brown; in all likelihood, it would have hurt him.