The intellectual functions of 64 epileptic patients who had had an initial evaluation between five and 16 years of age, including the WISC, were re-evaluated after a period of at least five years. In general the seizure states had improved, and 50 per cent were in remission for between two and eight years. All but four were still taking at least one anticonvulsant drug. WISC IQ estimates showed a slight decrease. Verbal and performance areas could be differentially affected, and a gain in one could be offset by a loss in the other, so the Full-scale IQ might not be a reliable measure of day-to-day performance. Those whose seizures remained uncontrolled had a statistically significant decrease in performance IQ, whereas in general it was stable or increased for patients in remission. There was evidence that decreased IQ indicated slower mental growth rather than loss of previously acquired function. Phenobarbital but not phenytoin levels were inversely correlated with IQ, suggesting that the upper limit of the 'therapeutic range' of phenobarbital may already be toxic with regard to learning abilities. To optimize an epileptic child's functioning in school and to prevent long-term intellectual problems, it is advisable that IQ testing should be part of the routine initial evaluation, and that drug levels should be checked at regular intervals.