William DAUBERT; Joyce Daubert, individually and as Guardians Ad Litem for Jason Daubert, a minor; Anita De Young, individually, and As Guardian Ad Litem for Eric Schuller, Plaintiffs-Appellants, v. MERRELL DOW PHARMACEUTICALS, INC., a Delaware corporation, Defendant-Appellee, 951 F.2d 1128

KOZINSKI, Circuit Judge. Plaintiffs Jason Daubert and Eric Schul-ler suffer from limb reduction birth defects. They allege that these defects resulted from the fact that their mothers used Bendectin, a prescription anti-nausea drug, during pregnancy. The plaintiffs seek damages from the drug’s manufacturer, defendant Merrell Dow Pharmaceuticals. Plaintiffs’ evidence of causation consisted primarily of expert opinion based on in vitro and in vivo animal tests, chemical structure analyses and the reanalysis of epidemiological studies.

Among the contrary evidence proffered by Merrell Dow was the affidavit of a physician and epidemiologist who reviewed all of the available literature on the subject, which included more than 30 published studies involving over 130,000 patients, and concluded that no published epidemiological study had demonstrated a statistically significant association between Bendectin and birth defects. Plaintiffs do not challenge this

summary of the published record. The district court determined that plaintiffs could not meet their burden of proving that Bendectin caused Jason’s and Eric’s birth defects and granted Merrell Dow’s motion for

summary judgment. Daubert v. Merrell Dow Pharmaceuticals, Inc., 727 F.Supp. 570 (S.D.Cal.1989). A. Expert opinion based on a scientific technique “is admissible if it is generally accepted as a reliable technique among the scientific community.” United States v. Solomon, 753 F.2d 1522, 1526 (9th Cir.1985) (citing Frye v. United States, 293 F. 1013, 1014 (D.C.Cir.1923)). We impose this requirement because such evidence “create[s] a substantial danger of undue prejudice or of confusing the issues or of misleading the jury ... because of its aura of special reliability and trustworthiness.” United States v.

Amaral, 488 F.2d 1148, 1152 (9th Cir.1973); see also Learned Hand, Historical and Practical Considerations Regarding Expert Testimony, 15 HarvL.Rev. 40, 50-52 (1901). Moreover, whether expert opinion can satisfy this requirement is reviewed de novo by an appellate court, for “[t]he answer to the question about the reliability of a scientific technique or process does not vary according to the circumstances of each case. It is therefore inappropriate to view this threshold question of reliability as a matter within each trial judge’s individual discretion.” Reed v.

State, 283 Md. 374, 391 A.2d 364, 367 (1978); see also United States v. Williams, 583 F.2d 1194, 1197-1201 (2d Cir.1978) (applying de novo review to the question of the general acceptance of a scientific technique), cert. denied, 439 U.S. 1117, 99 S.Ct. 1025, 59 L.Ed.2d 77 (1979). For expert opinion based on a given scientific methodology to be admissible, the methodology cannot diverge significantly from the procedures accepted by recognized authorities in the field.

If it does so diverge, it cannot be shown to be “generally accepted as a reliable technique,” Solomon, 753 F.2d at 1526, and a district court must exclude it. If such evidence is admitted and materially affects the verdict, a judgment supported by this evidence cannot stand. See Lindsey v. United States, 16 Alaska 268, 237 F.2d 893, 897-98 (9th Cir.1956). B. Whether Bendectin is responsible for limb reduction defects is not a new question in the federal courts.

Four of our sister circuits have considered whether plaintiffs could establish such responsibility in the absence of critically analyzed epidemiological studies establishing a connection between the use of the drug and the birth defects. DeLuca v. Merrell Dow Pharmaceuticals, Inc., 911 F.2d 941 (3d Cir.1990); Brock v. Merrell Dow Pharmaceuticals, Inc., 874 F.2d 307, modified, 884 F.2d 166 (5th Cir.1989), cert. denied, 494 U.S. 1046, 110 S.Ct. 1511, 108 L.Ed.2d 646 (1990); Richardson v.

Richardson-Merrell, Inc., 857 F.2d 823 (D.C.Cir.1988), cert. denied, 493 U.S. 882, 110 S.Ct. 218, 107 L.Ed.2d 171 (1989); Lynch v. Merrell-National Labs., 830 F.2d 1190 (1st Cir.1987) (opinion by Noonan, J., sitting by designation). Three circuits have concluded that the plaintiffs could not establish such a link.

Considering largely the same evidence presented to the district court below, these courts held that the animal and chemical studies were insufficient to establish a link between Bendectin and birth defects, Brock, 874 F.2d at 313-15; Richardson, 857 F.2d at 830; Lynch, 830 F.2d at 1194, in part because the plaintiffs’ own experts acknowledged the need to verify such data with epidemiological evidence. Brock, 874 F.2d at 314; Richardson, 857 F.2d at 830.

These courts were unwilling to allow plaintiffs to rely on reanalyses of epidemiological studies because these reanalyses had neither been published nor subjected to the rigors of peer review. They found the methodology particularly problematic in light of the massive weight of the original published studies supporting the defendant’s position, all of which had undergone full scrutiny from the scientific community. See Brock, 874 F.2d at 312-13; Richardson, 857 F.2d at 831; Lynch, 830 F.2d at 1193-96; see also Christophersen v. Allied-Signal Corp., 939 F.2d 1106, 1111 n. 8 (5th Cir.1991) (en banc).

Plaintiffs argue that reanalysis is a generally accepted scientific technique, so it follows that their experts were basing their opinions on a permissible methodology. But the reanalysis of epidemiological studies is generally accepted by the scientific community only when it is subjected to verification and scrutiny by others in the field. See Dore, A Commentary on the Use of Epidemiological Evidence in Demonstrating Cause4n-Fact, 7 Harv. Envtl. L.Rev. 429, 438-39 (1983) (discussing requirements for original epidemiological studies).

Plaintiffs’ reanalyses do not comply with this standard; they were unpublished, not subjected to the normal peer review process and generated solely for use in litigation. It does not suffice that the expert’s methodology meet some of the requirements imposed by the scientific community; it must meet all of the essential requirements. Selective borrowing from generally accepted scientific methodology does not satisfy Solomon’s rigorous standard. C.

Lynch, Richardson and Brock reflect a well-founded skepticism of the scientific value of the reanalysis methodology employed by plaintiffs’ experts; they recognize that “[t]he best test of certainty we have is good science — the science of publication, replication, and verification, the science of consensus and peer review.” P. Huber, Galileo’s Revenge: Junk Science in the Courtroom 228 (1991).

For the convincing reasons articulated by our sister circuits, we agree with the district court that the available animal and chemical studies, together with plaintiffs’ expert reanalysis of epidemiological studies, provide insufficient foundation to allow admission of expert testimony to the effect that Bendectin caused plaintiffs’ injuries. Accordingly, plaintiffs could not satisfy their burden of proving causation at trial.

See Christophersen, 939 F.2d at 1109 (“When the contested evidence is essential to the cause of action and the trial court has excluded the evidence, we may decide the appeal ... solely on the basis of the soundness of the evidentiary ruling. For if we uphold the exclusion of essential evidence, the [review of the

summary judgment ruling] becomes academic.”). AFFIRMED. . Richardson and Lynch held that expert testimony that Bendectin caused birth defects, based solely on animal and chemical tests and reana-lyses of epidemiological studies, was inadmissible under Federal Rule of Evidence 703, and therefore that plaintiffs had failed to present evidence to support a jury verdict in their favor. 857 F.2d at 830-831; 830 F.2d at 1193-96. Brock did not hold that the plaintiffs’ expert testimony was inadmissible, but did decide that it was insufficient to support the jury’s verdict. 874 F.2d at 315. .

The Third Circuit has left open the possibility that expert testimony based on the reanalysis of epidemiological studies may be admissible if it can be shown to be reliable and not too likely to overwhelm, confuse or mislead the jury. See DeLuca, 911 F.2d at 955 (remanding case for district court to reconsider the admissibility of plaintiffs’ expert testimony). Because the Third Circuit has specifically rejected the Frye/Solomon “generally accepted” standard, United States v.

Downing, 753 F.2d 1224, 1236-37 (3d Cir.1985), we find the court’s analysis in DeLuca to be contrary to our established precedent and decline to follow it. . Scientific studies conducted in anticipation of litigation must be scrutinized much more carefully than studies conducted in the normal course of scientific inquiry. This added dose of skepticism is warranted, in part, because studies generated especially for use in litigation are less likely to have been exposed to the normal peer review process, which is one of the hallmarks of reliable scientific investigation.

See Brock, 874 F.2d at 313; Richardson, 857 F.2d at 831 n. 55; Perry v. United States, 755 F.2d 888, 892 (11th Cir.1985).