We prospectively compared the activating effect of chlorpromazine (CHLP, 50 mg, i.m.) versus sleep deprivation (SD) in 41 patients with a clinical diagnosis of partial epilepsy. Patients were selected on the basis of both seizure onset in adulthood and normal interictal awake EEG recordings. Twenty out of 41 patients (group A) were not yet treated because of either recent onset of epilepsy or misdiagnosis. The remaining 21 patients (group B) were treated with antiepileptic drugs (AED). A control group consisted of 18 healthy, volunteers (group C). All EEG recordings were scored by one observer according to a fixed protocol. In group A, SD and CHLP activated sleep EEG in 12 (60%) and 19 (95%) patients, respectively. This difference reached the limit of statistical significance (P = 0.05, McNemar test). In group B, SD and CHLP activated sleep EEG in 12 (57%) and 13 (62%) patients respectively. There was a significant (P < 0.02, exact Fisher test) intergroup difference (95% vs 62%) with respect to the activating effect of CHLP. No false-positive results were found in 18 control subjects. SD or CHLP activating procedures did not provoke any epileptic seizures in any of the groups. In conclusion, EEG activation by either SD or CHLP is highly specific in the diagnosis of adult-on-set partial epilepsy. Moreover, CHLP is more sensitive than SD in untreated patients, whereas the activating effect of CHLP may be partially attenuated by AED.