Medication withdrawal is used in selected instances to precipitate seizures during evaluation of epileptic patients. Whether such medication manipulation can alter electrical seizure onset or clinical seizure type is not known. Information regarding the effects of anticonvulsant withdrawal on depth EEG onset and clinical manifestations of seizures was reviewed in 25 patients with partial complex seizures in whom intracranial electrodes were implanted. For the purposes of this study, the withdrawal period was defined as the time after abrupt cessation of medication between one and five half-lives of the drug and the base-line state as all other times. Ten patients were withdrawn from carbamazepine, 6 from phenobarbital, 2 from phenytoin, and 3 from primidone. Overall effects of anticonvulsant withdrawal in 18 patients and 21 instances of withdrawal with a total of 89 withdrawal and 71 base-line seizures were notable. Focal electrical onset and typical clinical seizure type during withdrawal were entirely consistent with electrical and/or clinical characteristics of seizures occurring during the base-line state in 13 instances. "New" clinical and/or electrical seizure types occurred during withdrawal only in a number of patients in whom bilateral or multifocal seizure onset was confirmed electrographically outside of the withdrawal period. In only 1 patient did withdrawal produce atypical clinical and electrical seizures when the base-line state showed a clinically typical seizure of clear-cut localized onset. After surgery this patient was shown to have unequivocal multifocal seizure onset. Thus, for purposes of localizing single epileptic foci, information obtained during the anticonvulsant withdrawal period provided no misleading information in this group of patients, while condensing the time needed for evaluation.