Opinion for the Court filed by Circuit Judge MIKVA.

MIKVA, Circuit Judge:

This appeal arises from a products liability action seeking recovery for a child’s birth defects allegedly caused by the mother’s use of Bendectin during her pregnancy. For over twenty-five years (1957-1983), Bendectin was a popular medication for morning sickness experienced during early pregnancy. In 1983, the manufacturer of Bendectin, Richardson-Merrell, Inc. (“Merrell”), voluntarily took the drug off the market although the Food and Drug Administration never rescinded approval of the drug. Bendectin has been studied extensively to determine whether it is a human teratogen, i.e. capable of causing birth defects in humans. To date, no published human population or epidemiological study has concluded that there is a statistically significant association between Bendectin and limb reduction defects of the type at issue in this case.

Merrell, the defendant-appellant in this action, appeals a jury’s award of $20 million in compensatory damages for plaintiff-appellees, the Ealys, and appeals the trial court’s order denying its motions for judgment n.o.v. and denying its motions for a new trial. The Ealys cross-appeal the trial court’s order granting a remittitur of the jury’s $75 million punitive damage award. Merrell argues that this court’s decision in *1160 Richardson v. Richardson-Merrell, Inc., 857 F.2d 823 (D.C.Cir.), cert. denied, - U.S. -, 110 S.Ct. 218, 107 L.Ed.2d 171 (1989) (“Richardson ”) requires reversal of the compensatory award. In addition, Mer-rell contends that the compensatory award is excessive and violates due process; and that numerous procedural errors regarding expert testimony and other evidence warrant a new trial. The Ealys counter each of these assertions, arguing, inter alia, that the Erie doctrine requires that this court follow the D.C. Court of Appeals decision in Oxendine v. Richardson-Merrell, Inc., 506 A.2d 1100 (D.C.1986) (“Oxen- dine I”).

We find that this case is squarely within the binding rule articulated in Richardson: an expert opinion that Bendectin is a human teratogen which caused the plaintiffs birth defects is without scientific foundation under Federal Rule of Evidence 703 in the face of “a wealth of published epidemiological data” to the contrary. 857 F.2d at 832. Accordingly such expert opinion is inadmissible. Because we discern no material difference between the evidence presented in Richardson and that presented in this case, we reverse the trial court’s denial of the motion for judgment n.o.v. Contrary to the Ealys’ assertions, neither this court nor the Richardson court deviated from the local tort law expressed in Oxendine I. Our decision in Richardson turned on the admissibility of expert testimony a procedural matter which does not implicate the principles underlying Erie Railroad Co. v. Tompkins, 304 U.S. 64, 58 S.Ct. 817, 82 L.Ed. 1188 (1938). It is also significant to a point of distinction that fewer scientific studies were introduced by the parties in Oxendine I than in either this case or the Richardson case.

Because wereverse on the issue of liabil- ity, we need not reach any of the other issues raised on appeal or cross-appeal. I. BACKGROUND

I. Background

sekou Ealywas born in 1979 with conge- nital birth defects of his hands and arms. He has three fingers on one hand and four on the other and no thumbs. His arms are underdeveloped and locked at the elbows. Although he is otherwise a normal child, he has undergone weekly psychiatric therapy to address his mental fixation that his par- ents cut off his fingers to punish him for being naughty. It is undisputed that Sek- ou’s mother took Bendectin during her pregnancy for morning sickness and that extensive evaluation has failed to disclose a clear cause for his malformations.

At trial,the plaintiffs argued that infer- ences drawn from analysis of Bendectin’s chemical composition, in vitro studies of the effects of Bendectin components on animal cells in test tube cultures, in vivo data from studies conducted on animals given Bendectin, and unpublished rein- terpretations of the epidemiological litera- ture are sufficient foundation to support an opinion that Bendectin is a human terato- gen. This was precisely the theory of proof used by the plaintiffs in Richardson. In developing their case, the Ealys intro- duced the testimony of eight expert wit- nesses, only one of whom did not testify in Richardson. This single witness, Dr. Wil- liam McBride, was also the only expert witness whose testimony, introduced by deposition, was based on published studies which he himself had conducted. Based on his study of the effects of doxylamine suc- cinate, an element of Bendectin, on rabbits and marmosets, Dr. McBride opined that Bendectin is a human teratogen.

After hearingthe expert testimony, the jury reached a verdict for the plaintiff, awarding $20 million in compensatory dam- ages and $75 million in punitive damages. The trial court rejected the defendant’s ar- gument, on motion for judgment n.o.v., that the verdict was without scientific foun- dation, concluding instead that there was “substantial conflicting evidence” about which reasonable people could differ. The court also noted that two other juries in similar Bendectin cases tried in the district court had reached verdicts for the plain- tiffs. One of those verdicts was later set aside in Richardson and the other is pend- ing decision on a post-trial motion to set aside the verdict on the authority of Rich- ardson. The trial court upheld the com- the com *1161 pensatory damages award, finding that the $20 million figure did not “shock the conscience” of the court. The court set aside the punitive damages award because there was no evidence of willful disregard or outrageous conduct by the manufacturer,

II. Discussion

A. Standard of Review

In Richardson, this court affirmed a judgment n.o.v. in favor of Merrell, holding that an expert opinion that Bendectin caused a child’s limb reduction defects was inadmissible based on the record developed in that case. 857 F.2d at 832. Appellant Merrell now argues that the Richardson decision provides a binding legal precedent that requires reversal of the trial court’s denial of its motion for judgment n.o.v. in the case sub judice. The standard of review of the trial court’s decisions on motions for judgment n.o.v. is de novo. As with motions for directed verdicts, the reviewing court asks the same question asked by the trial court: whether, viewing the evidence in the light most favorable to the plaintiff, and giving him every reasonable inference, the evidence is so one-sided that no reasonable juror could find on the admissible evidence that Bendectin caused the plaintiff’s birth defects. Richardson, 857 F.2d at 827-28; McNeal v. Hi-Lo Powered Scaffolding, Inc., 836 F.2d 637, 640-41 (D.C.Cir.1988). Any conflicts in evidence must be resolved in favor of the plaintiff who was the prevailing party below. McNeal, 836 F.2d at 641.

B. The Richardson Decision

The Richardson court concluded that there was not sufficient evidence in the record on which a jury could base a verdict for the plaintiff because all of the plaintiff’s expert testimony — the only relevant evidence on the causation issue — was inadmissible. Relying on Federal Rule of Evidence 703, the court ruled, in essence, that the issue of scientific causation should not have been left to the jury because Dr. Alan Done’s expert opinion that Bendectin caused the plaintiff's birth defects lacked an adequate scientific foundation. Dr. Done served as the lead expert for the plaintiffs in both Richardson and this case, The Richardson court’s treatment of Dr. Done’s testimony subsumed that of the other experts presented on the Richardsons’ behalf. 857 F.2d at 829 n. 38.

Rule 703 governs the “Bases of Opinion Testimony by Experts,” and requires that the grounds relied upon by an expert be of “a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject.” Fed.R. Evid. 703. The Richardson court noted that Dr. Done had predicated his opinion on four types of data. After canvassing the probative value of each type of data, the court concluded:

These three types of studies then— chemical, in vitro, and in vivo — cannot furnish a sufficient foundation for a conclusion that Bendectin caused the birth defects at issue in this case. Studies of this kind, singly or in combination, are not capable of proving causation in human beings in the face of the overwhelming body of contradictory epidemiological evidence. Perhaps mindful of this, the last type of evidence considered by Dr. Done consisted of the epidemiological studies. When such studies are available and relevant, and particularly when they are numerous and span a significant period of time, they assume a very important role in determinations of questions of causation.

857 F.2d at 830 (emphasis added). Every epidemiological study introduced in Richardson had concluded that there was not a statistically significant relationship between Bendectin and birth defects. The court proceeded to note Dr. Done’s concession that he would not “give an opinion as to causation without a statistically significant relationship.” Id. Dr. Done overcame this hurdle

[o]nly by recalculating the [published epidemiological] data ... to obtain what he deems a statistically significant result. Moreover, the studies rejected by Dr. Done had been published in peer-reviewed scientific journals, while Dr. Done *1162 has neither published his recalculations nor offered them for peer review.

Id. at 831.

The Richardson court distinguished its case from our decision in Ferebee v. Chevron Chemical Co., 736 F.2d 1529 (D.C.Cir.), ce rt. denied, 469 U.S. 1062, 105 S.Ct. 545, 83 L.Ed.2d 432 (1984), which addressed the question whether paraquat exposure caused pulmonary fibrosis. Ferebee, the Richardson court concluded, was an example of “ ‘a classic battle of the experts, a battle in which the jury must decide the victor.’ ” Richardson, 857 F.2d at 832 (quoting Ferebee, 736 F.2d at 1535). In Ferebee the defendant attempted to discredit the opinions of the plaintiffs experts on the ground that their views were sufficiently novel to be inadmissible. The Fere-bee court stated:

[A] cause-effect relationship need not be clearly established by animal or epidemiological studies before a doctor can testify that, in his opinion, such a relationship exists. As long as the basic methodology employed to reach such a conclusion is sound, ... products liability law does not preclude recovery until a “statistically significant” number of people have been injured or until science has had the time and resources to complete sophisticated laboratory studies of the chemical.

736 F.2d at 1535-36. The Richardson court found Ferebee inapposite because it applied to cases where the causation theory is novel and “ ‘stand[s] at the frontier of current medical and epidemiological inquiry.’ ” In Richardson, however, the case was a great distance from the frontier of “epidemiological inquiry,” because studies had been conducted and published for twenty-odd years. 857 F.2d at 832.

The Richardson court concluded, as a matter of law, that the “wealth of published epidemiological data ... none of which has concluded that the drug is teratogenic ... must be given their just due.” 857 F.2d at 832. Therefore, under Rule 703, an opinion refuting this scientific consensus is inadmissible for lack of an adequate foundation, in the absence of other substantial probative evidence on which to base this opinion. It is this uncontroversial rule of evidence that is the ratio decidendi of Richardson and this case.

Two other circuits have reached similar conclusions. See Lynch v. Merrell-National Laboratories, Inc., 830 F.2d 1190 (1st Cir.1987); Brock v. Merrell Dow Pharmaceuticals, Inc., 874 F.2d 307 (5th Cir.), modified, 884 F.2d 166 (5th Cir.), reh’g denied, 884 F.2d 167 (5th Cir.1989) (en banc). The First Circuit opined that “[a] new [epidemiological] study coming to a different conclusion and challenging the consensus would be admissible evidence. Without such a study there is nothing on which expert opinion on Bendectin as a cause may be based.” Lynch, 830 F.2d at 1194; cf. In Re “Agent Orange” Prod. Liab. Litig., 611 F.Supp. 1223, 1242 (E.D.N.Y.1985) (“When the expert’s ... testimony lie[s] at the periphery of what the scientific community considers acceptable, special care should be exercised in evaluating the reliability and probative worth of the proffered testimony under Rules 703 and 403”). Unlike the circumstances of Fere-bee, the body of published epidemiological opinions on the subject at hand is extensive, indeed massive, and all such opinions point to the same conclusion. As Richardson teaches, this is our measuring rod for scientific adequacy.

Because Richardson provides a binding legal precedent governing the admissibility of expert opinion on the ability of Bendec-tin to cause human birth defects, the Ealys can only avoid that decision by showing that the record here is materially different from that in Richardson. We find no such difference.

C. Evidentiary Differences

Merrell argues that there are no material differences between this case and Richardson because the Ealys have not pointed to any new epidemiological studies on Bendec-tin not introduced in Richardson. The Ea-lys do not refute that the epidemiological studies are identical in both cases. Here, as in Richardson, the plaintiff’s epidemiology expert, Dr. Shanna Swan, tried to re *1163 fute the validity of the published epidemiological data through her own unpublished reanalysis. The chemical structure, in vivo, and in vitro studies relied upon by the Ealys are also virtually the same as those referenced in Richardson with one exception: Dr. McBride testified by deposition as an expert for the Ealys but not for the plaintiffs in Richardson. He testified about his in vivo studies of the effects of doxylamine succinate on rabbits and marmosets. This data, however, along with all the other in vivo, in vitro, and chemical structure data, has dubious significance in the face of the epidemiological data.

The Ealys make no effort to point out differences in the scientific studies introduced in the two cases. Instead, they attempt to distinguish this case from that of Richardson by focusing on specific statements made by various experts on both sides that were not made in Richardson. None of these differences are material. According to Richardson, none of the evidence relied upon by the Ealys’ experts, either singly or in combination, provides an adequate scientific foundation to render the plaintiff experts’ opinions on human causation admissible. The Ealys offer nothing to alter this conclusion.

D. The Erie Doctrine

Equally unavailing is the Ealys’ contention that the Erie doctrine requires this court to follow the D.C. Court of Appeals decision in Oxendine I. That Bendectin case, which was litigated in the D.C. Superior Court, resulted in a verdict for the plaintiff. The trial judge then granted a judgment n.o.v. for Merrell, but the D.C. Court of Appeals reversed the trial judge and reinstated the jury verdict. Oxendine I, 506 A.2d 1100 (D.C.1986). The jury verdict was affirmed on other grounds in Oxendine v. Merrell-Dow Pharmaceuticals, Inc., 563 A.2d 330 (D.C.1989) (“Oxendine II”). The same four types of data relied upon by the Ealys were involved in the Oxendine ease. The court in Oxendine I conceded that each individual class of data showed “little or nothing when viewed separately from one another,” but reasoned that their sum formed a “mosaic” greater than its parts, providing “a foundation for Dr. Done’s opinion that Bendectin caused appellant’s birth defects.” Oxendine I, 506 A.2d at 1100.

The Richardson court specifically rejected the argument that it was required to follow Oxendine I because it was sitting in a diversity action on a state tort claim:

[T]he Richardsons’ Erie challenge must be rejected. We note that the Richard-sons’ argument ignores the differences in the evidence presented in the two cases. Moreover, the Erie doctrine applies to state substantive, not procedural, law to be followed by federal courts exercising diversity jurisdiction_ Because the admissibility of testimony, which is governed by the Federal Rules of Evidence, is the crux of our decision in this case ... the Richardsons’ argument must be rejected.

Richardson, 857 F.2d at 825 n. 9 (citations omitted). As our discussion thus far has shown, this case, like Richardson, turns on the admissibility of the expert testimony. Unfortunately for the Ealys, the only evidence they have on the issue of scientific causation is the testimony of their experts. Once that testimony is deemed inadmissible, it is self-evident that the evidence is insufficient to support a verdict for the plaintiffs. Contrary to the Ealys’ assertions, however, this court is enforcing a procedural rule governing admissibility; it is not supplanting substantive state rules on the sufficiency of evidence.

In any event, the Oxendine I court’s reasoning would not be directly exportable to the Ealys’ cause because, as with Richardson, there are material differences between the evidence in each case. Oxen-dine I was tried in 1983, Richardson in 1986, and the Ealys’ case in 1987. In the intervening years, two large epidemiological studies were published which found no association between Bendectin and birth defects.

III. Conclusion

It may well be that future scientific research will generate new epidemiological or *1164 other types of data providing an adequate basis for an expert’s opinion that Bendectin is a human teratogen that caused a plaintiffs birth defects. Until that day, the existing body of published epidemiological studies, all finding no significant statistical association between ingestion of Bendectin and birth defects, must be recognized as the measuring stick for the admissibility of expert testimony on this issue. On this narrow, procedural issue, this case is the same as Richardson. We reverse the trial judge’s denial of judgment n.o.v. because the expert opinion that Bendectin caused Sekou Ealy’s limb defects was without scientific foundation. We remand to the district court to enter a judgment n.o.v. for Merrell. This conclusion disposes of all other issues raised on appeal and cross-appeal.

So ordered.