Although some studies of cognitive and behavioral changes in patients with epilepsy may suggest an effect of AEDs, an unequivocal causal relationship is unlikely Epilepsia, Vol. 39, No. 9, 1998 to be established unless the study has been specifically designed to address such a causality. Various study de- signs have been used for the purpose of identifying and separating a possible drug effect, such as (a) studies in volunteers, (b) comparisons of the effects of two or more different drugs, (c) comparison of an AED versus pla- cebo as add-on drug, (d) comparison of performance at high and low levels of a drug, (e) testing patients before treatment and after the introduction of a drug in mono- therapy, and (f) testing seizure-free patients before and after discontinuation of a single drug. Each one of these methods has some shortcomings, but testing seizure-free patients before and after discontinuation of a mono- therapy may actually come closest to eliminating any other factors. Because observed deficits are likely to re- flect a possible effect of an AED superimposed on a background of preexisting deficits, studies in volunteers without epilepsy may not reflect conditions in patients. For instance, it is a common clinical observation that some patients with brain damage are more likely to dis- play neurotoxic adverse effects from AEDs. Studies in volunteers are also limited to short periods and can over- look either a cumulative long-term effect (underestima- tion) or the possible development of tolerance after a limited acute effect (overestimation); they will not be included in the present review. Considering that, in ad- dition to the differing study designs, psychological or behavioral test batteries or questionnaires differ among studies, it comes as no surprise that the literature on the subject contains many discrepancies.