OPINION
PER CURIAM:This case is before us on remand from the Supreme Court for the second time. We reiterate the facts and course of litigation very briefly; a fuller exposition may be found in our earlier opinions and orders that we cite here.
I
Shirley Ree Smith was convicted in California state court of assault on a child resulting in death. The state court of appeal affirmed her conviction, and the California Supreme Court denied review. Smith then filed a federal habeas petition claiming that her conviction violated due process because the evidence was constitutionally insufficient. The district court denied the petition and Smith appealed. We reversed and directed issuance of the writ. Smith v. Mitchell, 437 F.3d 884 (9th Cir. 2006). We held that no rational trier of fact could have found beyond a reasonable doubt that Smith caused the child’s death, and that the state court’s affirmance of Smith’s conviction was an unreasonable application of Jackson v. Virginia, 443 U.S. 307, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979). Smith, 437 F.3d at 890.
The State’s petition for panel and en banc rehearing was denied, Smith v. Mitchell, 453 F.3d 1203 (9th Cir.2006), and the State filed a petition for certiorari in the Supreme Court. The Supreme Court granted certiorari, vacated our decision, and remanded for further consideration in light of its intervening decision in Carey v. Musladin, 549 U.S. 70, 127 S.Ct. 649, 166 L.Ed.2d 482 (2006). Patrick v. Smith, 550 U.S. 915, 127 S.Ct. 2126, 167 L.Ed.2d 861 (2007).
On remand, we held that Carey v. Musladin did not cast any doubt on our earlier conclusion that Smith’s case fell squarely within Jackson, and that the state court’s denial of her claim of constitutionally insufficient evidence was an unreasonable application of Jackson. 508 F.3d 1256, 1261. We also held that a later intervening Supreme Court decision in Schriro v. Landrigan, 550 U.S. 465, 127 S.Ct. 1933, 167 L.Ed.2d 836 (2007), did not affect our earlier result. Id. at 1260. We accordingly reinstated our previous opinion and judgment that a writ of habeas corpus must issue. Id. at 1261.
*1237We subsequently denied the State’s petition for panel and en banc rehearing, rejecting the State’s contention that the Supreme Court’s recent decision in Wright v. Van Patten, 552 U.S. 120, 128 S.Ct. 743, 169 L.Ed.2d 583 (2008), required us to reverse course.
The State once again petitioned for certiorari. The Supreme Court granted the writ, vacated our judgment, and remanded for further consideration in light of McDaniel v. Brown, — U.S. -, 130 S.Ct. 665, 175 L.Ed.2d 582 (2010) (“Brown ”).
We have now examined Brown along with supplemental briefs from the parties addressing its potential effect on Smith’s case. We conclude that nothing in Brown is inconsistent with our prior decision or our method of reaching it. We accordingly reinstate our former decision, reported at 437 F.3d 884.
II
Brown involved a decision of this court that, like our decision in Smith, held that a state conviction failed to pass constitutional muster under the standard of Jackson v. Virginia. There the similarity of the two cases ends.
Brown’s conviction rested partly, but not entirely, on DNA evidence. In presenting that evidence, the prosecution’s expert introduced inaccuracies by testimony “equating random match probability with source probability, and [by] an underestimate of the likelihood that one of [the defendant’s] brothers would also match the DNA left at the [crime] scene.” 130 S.Ct. at 671. In applying the Jackson standard, this court took into account an expert report (“the Mueller Report”) commissioned by the defense for its collateral attack some 11 years after the trial. By the time the case reached the Supreme Court, Brown had conceded that this consideration of posttrial evidence was improper because the Jackson standard addresses whether any rational jury could convict on the evidence presented at trial. The Supreme Court stated that the concession was clearly correct. Id. at 672.
The Supreme Court went on, however, to rule that, even if the Mueller Report could have been considered, it did not render the evidence insufficient because it did not dispute that the DNA evidence matched the defendant. “That DNA evidence remains powerful inculpatory evidence even though the State concedes [its expert] overstated its probative value.... Even under Mueller’s odds, a rational jury could consider the DNA evidence to be powerful evidence of guilt.” Id. at 673.
The Supreme Court also faulted this court for treating inconsistencies in factual testimony (such as the time the defendant left a bar on the night of the crime, or the reason that he washed his clothes when he later arrived home) in a manner indicating that it had failed to review the evidence in the light most favorable to the prosecution. See id. The Supreme Court stated that this court “further erred in finding that the Nevada Supreme Court’s resolution of the Jackson claim was objectively unreasonable.” Id. at 674.
Ill
Our decision in Smith did not share the deficiencies that the Supreme Court pointed out in our opinion in Brown. First, and perhaps most important, the record in Smith’s case did not contain “powerful evidence of guilt,” Brown, 130 S.Ct. at 673, that a rational jury could accept as proof of guilt beyond a reasonable doubt. Smith was convicted on the theory, presented by the prosecution experts, that she had shaken her seven-week-old grandchild to death while the child’s mother slept some twenty feet away. There was no dispute in the evidence that Smith theretofore had been *1238a caring grandmother for the three children commonly left in her care. No one saw her shake the baby with the violence that the prosecution witnesses said must have caused the death. Indeed, the emergency personnel and emergency room physicians responding to the 911 call of Smith and her daughter considered the death to be an instance of Sudden Infant Death Syndrome, and that diagnosis changed only when the autopsy revealed some blood, both old and new, in the victim’s brain. We described the evidence in our decision:
All of the expert testimony offered by both sides agreed that the amount of recent bleeding (approximately one or two tablespoons) was not sufficient to have caused death, nor was the small abrasion sufficient for that purpose. To the prosecution experts, however, the presence of blood supported the diagnosis of Shaken Baby Syndrome. There was no dispute, however, that the usual Shaken Baby Syndrome death occurs from massive bleeding or swelling of brain tissue that creates such crushing pressure against the brain stem that vital processes are interrupted and the baby dies. It was also agreed that in 80% or more of the cases of Shaken Baby Syndrome, there is bleeding in the retinas of the eyes. There are also frequently fractures in the arms or similar evidences of violence.
There was no swelling, and only a small, non-fatal amount of bleeding, in Etzel’s case. Etzel had no retinal bleeding, and no fractures or large bodily bruises common in cases of shaking. The scalp abrasion was minimal, and was not even discovered until well into the autopsy. The prosecution experts testified, however, that shaking caused the death even though the physical examination of the brain during and after autopsy could not demonstrate that fact. The experts testified that the shaking must have been so violent and severe that it directly tore or sheared parts of the brain stem, causing immediate cessation of vital activity such as breathing. This tear in the brain stem would not have been apparent in autopsy, according to the prosecution experts, because instantaneous death would have prevented any bleeding or swelling. No microscopic examination of the brainstem was performed following the autopsy because, as Dr. Erlich testified, “[W]e wouldn’t have seen anything anyway.” The fatal tear or shearing would not have been detectable. Dr. Ehrlich could not identify any source in the literature for her hypothesis of undetectable brain stem shearing, but said she had learned it from lectures and consultations.
437 F.3d at 887 (footnote omitted). Thus the theory that Smith had shaken the baby so violently that the brain stem sheared in a way that caused immediate death with no bruising or bleeding could not be verified by examination of the brain. Nothing in the physical evidence supported the prosecution experts’ testimony as to the cause of death. We concluded that this evidence was “simply not the stuff from which guilt beyond a reasonable doubt can be established, especially in the face of all the other circumstances, many of which were recited by the magistrate judge, making the crime unlikely.” Id. at 890.
The state of the trial record in this case, therefore, is entirely different in degree and kind from that in Brown. And there is no question that, in Brown, the Supreme Court relied heavily on the evidence of guilt that remained for the jury even if the expert’s inaccuracies were combed out. Indeed, in introducing its holding in Brown, the Court stated:
We granted certiorari to consider whether [the district and appeals] courts misapplied Jackson. Because the trial *1239record, includes both the DNA evidence and other convincing evidence of guilt, we conclude that they did.
130 S.Ct. at 667 (emphasis added). We are satisfied that no comparable statement can be made about the state of the record in Smith.
Next, it almost goes without saying that our decision was not subject to the error that was conceded by the defense in Brown: consideration of evidence that was not before the jury. Our decision in Smith that no rational jury could find Smith guilty beyond a reasonable doubt was based entirely on the evidence presented to the jury at trial. That aspect of Broum is inapplicable here.
Finally, we conclude that our decision in Smith did not fail to accept the prosecution’s view of any evidence of historical fact, thus failing to accord deference to the state court’s application of Jackson. See Brown, 130 S.Ct. at 673. In truth, there was virtually no conflict in the evidence of the historical facts in Smith. The testimony regarding Smith’s loving care of her grandchildren was uneontroverted. The testimony regarding events on the night of the victim’s death was also not in conflict. There was not even any dispute over the condition of the victim’s body immediately after death and later during the autopsy. There was no dispute over the usual indications of Shaken Baby Syndrome, such as extensive bruising or broken limbs, detached retinas, and massive bleeding and swelling of the brainstem, all of which were absent in this case. The only dispute rose from the prosecution’s expert testimony that Smith must have shaken the baby so violently that it sheared the brainstem in a way that was undetectable because death was instantaneous. In sum, there was no verifiable evidence to support the prosecution experts’ testimony as to the cause of death. In concluding that no rational jury could find beyond a reasonable doubt, in light of all the evidence, that Smith had shaken the baby to death, we did not resolve any disputes of historical fact against the prosecution. We simply assessed the prosecution’s evidence on its own terms and concluded that it did not meet the Jackson standard, and that it was so lacking that the state court’s rejection of Smith’s argument over the insufficiency of the evidence was an unreasonable application of Jackson to the facts of this case.1 We do not find anything in Brown to cast doubt on these rulings.
As we indicated in our original opinion, we are acutely aware of the double layer of deference required by the Jackson standard when it is combined with the standard of the Antiterrorism and Effective Death Penalty Act, 28 U.S.C. § 2254(d). See Smith, 437 F.3d at 888-89; cf. Renico v. Lett, — U.S. -, & n. 1, 130 S.Ct. 1855, 1862 & n. 1, 176 L.Ed.2d 678 (2010). We acknowledged the double layer of deference in holding that, even with it, the California Court of Appeal had unreasonably applied Jackson in finding the evidence constitutionally sufficient. See id. at 890. We did not lightly conclude that, “[i]n this most unusual ease,” id. at 889, the writ must be granted even after the requisite deference is accorded to the decisions of the jury and the reviewing state court. Cases of constitutional insufficiency of evidence where the writ must issue even after consideration of the double level of deference will necessarily be rare, confined to extraordinary cases. We are satisfied that this is such an extraordinary case, and *1240that “[a]s a result of the unreasonable application of Jackson, there has very likely been a miscarriage of justice.” Smith, 437 F.3d at 890.
IY
Pursuant to the Supreme Court’s mandate, we have further considered our decision in light of the Supreme Court’s decision in McDaniel v. Brown. For the reasons set forth above, and in light of our prior opinions and orders cited in Section I of this opinion, we conclude that Brown does not cast doubt on the correctness of our decision, of which we remain convinced. We accordingly reinstate our earlier judgment and opinion, as reported at 437 F.3d 884.
OPINION AND JUDGMENT REINSTATED.
. We recognize that, to permit relief by habeas corpus, the state court’s application of Jackson to the facts of this case must be " 'objectively unreasonable.' ” Brown, 130 S.Ct. at 673 (quoting Williams v. Taylor, 529 U.S. 362, 409, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000)). We are satisfied by the state of the evidence that this requirement was met.