ORDER

The panel voted to deny the petition for panel rehearing. Judge Pregerson has voted to deny the petition for rehearing en banc, and Judges Canby and Reed have so recommended.

The petition for en banc rehearing has been circulated to the full court. A judge requested a vote on whether to rehear the matter en banc. The matter failed to receive a majority of the votes of the nonrecused active judges in favor of en banc consideration. Fed RApp. P. 35.

The petition for panel rehearing and the petition for rehearing en banc are denied.

I write to make clear our court has, by its decision in this case, made a substantial departure from settled principles of review of jury determinations of fact in criminal cases.

In this case, our court decides the opinions of three Board-certified physicians called by the prosecution that Smith’s shaking of baby Etzel caused his death must be substituted with the contrary opinions of non-Board-certified physicians called by the defense. Why? Because the defense’s doctors testified that a finding was absent on autopsy, and that finding was crucial and undermined the prosecution experts’ testimony. The three physicians called by the prosecution disagreed with the defense doctors, and explained why such a finding was not crucial. Our court simply accepts the defense theory and rejects the prosecution’s evidence. The jury was perfectly able to do just that. But when our court does it, it steps over the line dividing the province of the jury from that of the court.

This decision would be bad enough were we reviewing a district court’s judgment. But here, it is doubly bad for we are reviewing a state court decision under the Antiterrorism and Effective Death Penalty Act (“AEDPA”) which severely restricts the scope of our review, and mandates that “we apply the standards of Jackson with an additional layer of deference.” Juan H. v. Allen, 408 F.3d 1262, 1274 (9th Cir. 2005), cert. den., — U.S. ——, 126 S.Ct. 1142 , 163 L.Ed.2d 1000 (2006).

I

In 1997, Shirley Ree Smith was convicted by a California jury of the unlawful killing of her seven-week-old grandson, Etzel Glass (“Etzel”). On the night of Etzel’s death, Smith was staying with her daughter Tomeka, the infant’s mother, along with her two other young grandchildren and two of Tomeka’s sister’s children. Around midnight, Tomeka fed, changed, and washed baby Etzel before placing him to sleep on the living room sofa. Baby Etzel shared this sofa with his 18-month-old brother, Yondale, and four-year-old sister, Yolanda. Smith was also in the living room, sleeping on the floor. When Tomeka put Etzel to sleep at 11:30 p.m., he appeared healthy.

In her two interviews, Smith recounted different versions of the events that followed. First, Smith told a social worker, Linda Reusser, that she awoke after 3 a.m. when Etzel’s brother Yondale had a nightmare. After comforting Yondale, Smith went over to check Etzel. Etzel didn’t respond to her touch; she picked Etzel up and his head “flopped back.” She then gave Etzel “a little shake, a jostle to awaken him,” to which Etzel did not respond. Reusser described the shaking to the jury as “a quick jostle,” a “smooth motion.”

At this point, Smith stopped speaking. When Reusser prompted Smith to continue, she said “something like ‘Oh, my God. Did I do it? Did I do it? Oh, my God.’ ” Smith’s daughter Tomeka turned to Smith and said: “If it wasn’t for you, this wouldn’t have happened.” Smith didn’t say anything.

Smith told a slightly different story to the police who interviewed her as part of the criminal investigation. Smith stated that Yondale awoke from his nightmare sometime before 3 a.m.; Smith rose and checked Etzel; Etzel was fine. Then, past 3 a.m., Etzel’s sister Yolanda rolled off the couch and fell onto Smith. On waking, Smith noticed that Etzel’s diaper needed changing. After going to the bathroom to take her medicine, Smith picked up Etzel and saw he had “spit up” around his mouth, and his head was “flopped back.” Smith said something to Etzel and he didn’t respond; he was not breathing or moving. At first, Smith told the police she “shook” Etzel, but then corrected herself, and said she “twisted” him back and forth to get a response. When asked about her statement to Reusser, Smith denied saying that she had “shaken” Etzel.

Smith then carried Etzel, who was not responsive, into Tomeka’s room. Smith and Tomeka called 9-1-1. After unsuccessful attempts at cardio-pulmonary resuscitation (“CPR”) by the family and the paramedics, Etzel was taken to the hospital. Soon after his arrival, he was declared dead. The physician attending at the hospital suspected Etzel had died of Sudden Infant Death Syndrome (“S.I.D.S.”), which, as one defense expert put it, is “a medical[ly] sophisticated way of saying the child died and we really don’t have any idea what it died from.” The only injury the paramedics noticed was fresh blood in one of Etzel’s nostrils. In such cases, the doctor lists “suspected S.I.D.S.” as the cause of death, pending an autopsy.

But here, the autopsy revealed signs of recent trauma to Etzel’s brain. When the autopsy surgeon lifted Etzel’s brain out of the skull, she saw fresh blood on top of the brain (“subdural” blood). The subdural blood measured one or two tablespoons. The surgeon also saw a fresh blood clot between the hemispheres of Etzel’s brain, and recent hemorrhaging around the optic nerves. Further, she found a small quantity of fresh “subarachnoid” blood. Finally, the surgeon and her supervisor noticed a small bruise at the left lower-back part of Etzel’s head, and a recent abrasion at the same site.

Dr. Carpenter, the autopsy supervisor, opined Etzel had died by being violently shaken (“shaken baby syndrome”). According to Carpenter, death from violent shaking can occur in three ways: (1) massive swelling of the brain; (2) massive bleeding sufficient to crush the brain stem; or (3) a sudden shaking “so violent that it destroys the vital centers in the brain and is a quick death.” Here, Carpenter opined that death occurred through the last process, as Etzel’s head had undergone whiplash from chin to chest. The death occurred too quickly for visible trauma to develop on the brain stem itself.

Dr. Carpenter explained the basis of his opinion as the recent trauma to Etzel’s brain and the absence of other causes. The subdural blood, the subarachnoid blood, and the blood around the optic nerves showed “violent trauma to the head sufficient to cause the death of the infant.” The bruise and abrasion had, in Carpenter’s opinion, “very probably” occurred during the shaking episode, as the head collided with a hard, rough surface.

The alternate causes of death that Carpenter considered didn’t make sense to him. First, Carpenter ruled out S.I.D.S. because “[a case] is never called a S.I.D.S. if there is any suspicion of trauma to the infant.” Second, Carpenter ruled out that a fall from the sofa onto a carpeted floor could have caused Etzel’s injuries or death. Third, Carpenter considered the evidence of an old injury to Etzel’s brain, such as birth trauma, as the cause of death. While he could not exclude the old injury as contributing to Etzel’s death, the old injury was “not sufficient to cause death in that the infant had apparently sufficiently recovered from this injury and was appearing to look normal to others.”

Under Jackson, Dr. Carpenter’s opinion alone suffices to support the jury’s finding on causation. But then the prosecution called two other Board-certified doctors, who testified that violent shaking was the cause of death. Notably, the prosecution’s witness on rebuttal, Dr. Chadwick, had published articles on how to distinguish between falls and abusive injuries in children. Based in part on his own research, Chadwick opined that a fall from the sofa was “extremely unlikely” to cause death. Chadwick also explained that the old trauma was not the cause of death because old injuries do not cause sudden death without a specific pathology that was absent in this case. In short, Chadwick saw “no other natural or unnatural cause except the injury that would explain [Etzel’s] death,” and, therefore, opined that Etzel died from shaken baby syndrome.

The defense’s two expert witnesses, by contrast, opined that the necessary physi cal evidence of shaken baby syndrome was lacking. Dr. William Goldie opined the cause of death was S.I.D.S. Dr. Richard Siegler opined that the death was traumatic, but that it was impossible to isolate the cause of death between the recent and the old trauma. Significantly, both Goldie and Siegler would only diagnose Etzel’s death as shaken baby syndrome on a finding of visible injury to the shorn region of the brain stem. Even if death were instantaneous, Goldie opined that hemorrhages in the neck or brain stem would be present. Thus, the experts for the prosecution and the defense disagreed on a fundamental point: to be valid, does a doctor’s opinion that a baby died from violent shaking require evidence, visible on autopsy, of brain stem shearing?

The jury resolved this conflict among the experts’ opinions against Smith. The California Court of Appeal, after reviewing the medical evidence, affirmed the conviction because “[i]t was for the jury to resolve the conflicts” in that evidence. The California Supreme Court denied review. On federal habeas review, the Magistrate Judge recommended denying Smith’s petition for the same reason as the California Court of Appeal. Notwithstanding the tragedy of Smith’s case, it was the jury’s province to choose between the conflicting expert opinions. The district court accepted the recommendation and denied Smith’s habeas petition.

II

Our court, however, granted habeas relief on the ground of insufficient evidence of causation. See Smith v. Mitchell, 437 F.3d 884 (9th Cir.2006). Claims of insufficient evidence are judged according to a familiar standard: we ask whether, “after viewing the evidence in the light most favorable to the prosecution, any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.” Jackson, 443 U.S. at 319, 99 S.Ct. 2781 . AEDPA requires the federal courts to review Jackson claims with additional deference, and only grant habeas relief where the state court’s adjudication of a Jackson claim is objectively unreasonable. See Smith, 437 F.3d at 889 (citing Juan H., 408 F.3d at 1274 -75 & nn. 12-13).

To grant habeas relief, the opinion set aside the qualified opinions of the prosecution’s experts by misconstruing the basis for their opinions. “[Tjheir testimony was that death was caused by the shearing or tearing of the brain stem and” — according to the opinion — “they reached this conclusion because there was no evidence in the brain itself of the cause of death.” Smith, 437 F.3d at 890 .

Not so; the opinion is inaccurate.

The physicians called by the prosecution reached their conclusion despite the lack of visible shearing, not because of it, and explained why. Indeed, what provided the basis for the doctors’ opinions was the evidence of recent trauma to Etzel’s- brain: (1) the subdural hemorrhaging; (2) the subarachnoid hemorrhaging; (3) the hemorrhaging around the optic nerves; (4) the blood clot between the hemispheres of Etzel’s brain; and (5) the bruise and abrasion at the lower back of Etzel’s head. The prosecution’s experts considered and rejected other causes of Etzel’s death, such as a fall from the sofa, or the aggravation of Etzel’s older injury. Since none of these alternate theories explained Etzel’s death, the prosecution’s doctors opined that Etzel died from violent shaking, as evidenced by the trauma.

Not once did the prosecution’s doctors voice the supposed basis the opinion attributes to them. Rather their opinions were based on the evidence of the child’s injuries, notwithstanding that they did not see the brain stem shearing on autopsy.

In effect, the opinion adopts the defense experts’ view of what physical evidence is necessary to support a valid diagnosis of shaken baby syndrome. Specifically, the opinion relies on Dr. Siegler’s aspersion of the prosecution’s theory as “total fantasy,” when there is “no way to confirm it or deny it” with direct physical evidence. The opinion similarly concludes “there simply was no evidence to permit an expert conclusion one way or the other,” absent the physical evidence Siegler demanded. See Smith, 437 F.3d at 890 .

But the prosecution’s experts explained why they would not expect to see physical evidence of tearing in Etzel’s brain stem: Etzel’s death happened so quickly that the effects of the trauma did not have time to develop. That is, there was no swelling in the torn brain stem tissue because the child’s instant death closed down its circulation.

The opinion buttresses its conclusion by pointing to the absence of various factors that are “typically,” “usually,” or “frequently” present in cases of shaken baby death. See Smith, 437 F.3d at 887 . For instance, the opinion points to the absence of retinal bleeding, which is present in 80-85% of shaken baby cases, and almost never present otherwise. Yet Dr. Carpenter testified that the absence of retinal bleeding did not change his opinion on the cause of death. Even the defense doctor, Dr. Siegler, did not think retinal hemorrhages were necessary to diagnose shaken baby syndrome. Thus, the absence of these “usual” symptoms simply constitutes circumstantial evidence favorable to the defense, not a failure of proof as to an essential element of the prosecution’s case.

Although Smith did not challenge the admissibility of the prosecution’s experts’ testimony, the opinion’s analysis effectively makes the qualification of experts an issue on appeal. Up to now, the ceaselessly litigated question of whether expert testimony is “junk science” has been channeled through the Daubert and Kelly/Frye tests governing the qualification and admissibility of expert testimony. Under our court’s approach, a federal court of appeals may, effectively, set aside an expert opinion where it conflicts with the views of the other side’s experts.

Here, the prosecution’s experts based their opinions on the evidence of recent trauma to Etzel’s brain, and explained how a rapid death would result in brain-stem tearing that could not be seen. When the defense’s experts disputed the validity of this hypothesis, it was for the jury to resolve the conflicting opinions. Accordingly, the California Court of Appeal coi'rectly applied Jackson in affirming Smith’s conviction.

Because our court’s rejection of qualified expert opinions distorts Jackson analysis and contravenes AEDPA’s required deference, I respectfully dissent from the denial of rehearing en bane.

. Pub.L. No. 104-132, 110 Stat. 1214(codified as amended at 28 U.S.C. § 2241 et seq.) . As an expert testified below, the brain is protected by three sheaths or membranes: the dura, the arachnoid, and the pia. Blood found between the dura and the arachnoid is referred to as "subdural.” Blood between the arachnoid and the pia is referred to as "sub-arachnoid.” . Dr. Carpenter was Board-certified in forensic, anatomic, and clinical pathology. Tr. 534-36. "Board-certified” means the physician has first practiced a certain number of years in the field of specialty which the Board regulates. To gain certification as a specialist by the Board, the physician must then take certain written tests. Last, upon successful completion of those tests, the physician must sit successfully for oral examination by a group of Board-certified specialists. Only then can the physician become "Board-certified.” . While Dr. Carpenter could not rule out that Etzel’s head received a blow from a weapon, the bruise and abrasion were too small to indicate a blow that could have caused all of the trauma in Etzel's brain. .The overlap between the sufficiency of the evidence and qualification of experts is illustrated by the opinion's reliance on United States v. Boissoneault, 926 F.2d 230, 234 (2d Cir.1991). There, the Second Circuit reversed a conviction for possession of cocaine with the intent to distribute. The defendant was found carrying 5.31 grams of cocaine; $1,460 in ten and twenty dollar bills; and many slips of paper bearing names, addresses, and telephone numbers. Id. at 231 . The question was whether this evidence permitted the inference of intent to distribute the 5.31 grams. Over defense objection, the government called an expert who opined that the evidence was consistent with "classic street level distribution of small quantities of cocaine.” Id. at 232 . On appeal, the defendant challenged both the admission of this testimony and the sufficiency of the evidence. Id. at 231 . The Second Circuit reversed for insufficient evidence. Without deciding the evidentiary issue, the court held that conclusory opinions that "ambiguous conduct constitutes criminal activity” do not count toward establishing the sufficiency of the evidence. Id. at 234 . Boissoneault is directly contrary to California Evidence Code § 805, which allows opinion testimony as to ultimate facts. Moreover, the subject of the opinion in Boissoneault was whether ambiguous conduct is criminal, an area where expert opinion is of little help to the trier of fact. See United States v. Young, 745 F.2d 733, 765-66 (2d Cir.1984) (Newman, J., concurring). By contrast, whether autopsy results establish the cause of death is a question on which expert opinion is indispensable to the jury. . Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 593 , 113 S.Ct. 2786 , 125 L.Ed.2d 469 (1993). . People v. Kelly, 17 Cal.3d 24 , 130 Cal.Rptr. 144 , 549 P.2d 1240, 1244 (1976) (citing Frye v. United States, 293 F. 1013, 1014 (D.C.Cir.1923)).