A higher rate of congenital anomalies has been found after prenatal exposure to some combinations of antiepileptic drugs than to the separate drugs. In an earlier study a rate of 58% congenital anomalies was found among infants exposed to carbamazepine plus phenobarbitone plus valproate. In this study an attempt was made to determine whether this specific combination of drugs has teratogenic activity due to metabolic interaction. The epidemiological data were analyzed further. The high rate of congenital anomalies after prenatal exposure to this combination could not be explained by the effects of one or two of these drugs only, nor by additional exposure to phenytoin. Assuming that metabolic interaction in the arene oxide pathway resulting in accumulation of epoxide intermediates of antiepileptic drugs could be responsible for teratogenesis, the ratio of carbamazepine to carbamazepine-10, 11-epoxide concentrations in serum was determined in adult patients with epilepsy who were treated with carbamazepine only and with different combinations of phenobarbitone, valproate, and/or phenytoin. For carbamazepine monotherapy the mean ratio was 8.19. For all combinations lower ratios were found, indicating accumulation of carbamazepine-10,11-epoxide. The combination of carbamazepine, phenobarbitone, valproate, and phenytoin showed the lowest ratio (1.94), followed by carbamazepine, valproate, and phenytoin and by carbamazepine, phenobarbitone, and valproate (2.81 and 3.18, respectively). These results give rise to the question of whether the combination of carbamazepine, phenobarbitone, valproate, and/or phenytoin has teratogenic activity by accumulation of carbamazepine-10,11-epoxide or other epoxide intermediates, and stress the need to take metabolic interactions into account when investigating the teratogenic activity of antiepileptic drugs.