Data on clinical teratogenicity are at best derived from carefully conducted observational studies, whereas randomized, controlled trials have no place in this research area. We can only expect level B recommendations and lower. New relevant information has become available during the last 5 years on pregnancy outcomes with 3 of the most frequently used AEDs: carbamazepine, valproate, and lamotrigine. It seems that birth defect rates with arbamazepine monotherapy are lower than previously thought. In some large studies rates are only marginally increased compared with different control populations. More recent data do not suggest adverse effects of carbamazepine on cognitive development. The overall prevalence of malformations in association with lamotrigine exposure seems to be similar to that of carbamazepine. The only available prospective study on cognition does not indicate any adverse effects of lamotrigine. Malformation rates with valproate have consistently been found to be 2 to 3 times higher compared with carbamazepine or lamotrigine. More limited data also suggest adverse effects of high doses of valproate on cognitive development of the exposed child. For newer generation AEDs other than lamotrigine, data are still too limited to determine the risks for birth defects and are nonexisting with respect to possible adverse effects on cognitive development. Doses are important, and evidence is lacking for higher risks with valproate compared with other AEDs if doses are less than 800 to 1000 mg/d. Confounding factors contribute to some of the apparent differences between AEDs in pregnancy outcomes, and more data are needed, particularly concerning cognitive outcomes and specific birth defects. Based on these observations, valproate should not be a first-line AED for women who are considering pregnancy. In this situation this drug is best avoided if other effective but safer AEDs can be found for each individual woman's seizure disorder. Based on pregnancy outcome data, carbamazepine seems comparatively safe and a reasonable first-line choice in localization-related epilepsy. Alternatives are less clear in idiopathic generalized epilepsies. Lamotrigine seems comparatively safe, but its use in pregnancy is complicated by pharmacokinetic changes and risks of breakthrough seizures. The experience with use of levetiracetam and topiramate during pregnancy is still insufficient. Any attempt to change drugs should be completed and evaluated before conception; withdrawals or other major changes should be avoided during pregnancy. These conclusions are largely in line with the recently published report of the Quality Standards Subcommittee and Therapeutics and Technology Subcommittee of the American Academy of Neurology and the American Epilepsy Society.