Evidence accumulated over the past three decades has established AEDs as human teratogens. Important developments in the delineation of these compounds as human teratogens include: the demonstration of a consistent association between in utero exposure to AEDs and an increased occurrence of single major malformations, the description of AED-induced dysmorphogenic syndromes; demonstration of a dose-response relationship, both in terms of the number and dosage of AEDs; and evidence that pharmacogenetic differences in the metabolism of AEDs are strongly correlated with the occurrence of congenital malformations. Furthermore, the experimental animal findings, having accumulated in parallel to those of human studies, strongly support the teratogenic role of AEDs. Areas that require further amplification and clarification in future studies are the relative contribution of AEDs and other factors, such as genetic predisposition and maternal seizures, particularly with respect to the occurrence of minor anomalies, growth retardation, and developmental outcome; the relative teratogenicity of specific monotherapies and polytherapies; the predictive role of pharmacogenetic differences in the metabolism of AEDs in the occurrence of structural and functional abnormalities; and characterization of the precise nature of the pharmacogenetic defect underlying the aforementioned differences in AED metabolism. Attempts should also be made in future prospective studies to monitor metabolite levels of AEDs, particularly the oxidative metabolites, in order to further elucidate the relative contribution of individual differences in metabolism in the determination of adverse fetal outcome. Similarly, further efforts should be made to assess the clinical significance of decreased growth parameters in terms of mental and neurologic development, and to ascertain whether there is any risk for such abnormalities in children who do not display overt or persistent reductions in physical growth parameters. This is critically important in light of the animal studies that have shown functional abnormalities at doses that do not necessarily produce structural defects. Future investigations would be conducted through collaborative studies that would encompass sufficiently large numbers of women to provide adequate power to the statistical analyses of the data obtained. Care would have to be exercised to establish a uniform protocol for the collaborating centers. Regionally based investigations would be preferable to studies based at special centers, in order to assess the relative role of risk factors associated with abnormal pregnancy outcomes in the epileptic population at large.(ABSTRACT TRUNCATED AT 400 WORDS)