For 18 previously untreated patients with absence, myoclonic, or grand mal seizures--or combinations--results of clinical and electroencephalographical monitoring are reported. Sodium valproate was given once daily in the evening. Monitoring included repeated 24-48 h EEG recordings and drug blood level measurements. Results indicated the evening monodose to be an adequate therapeutic schedule for a considerable number of patients. Apart from the aspect of simplification, a further important aim is to individually minimize the drug dose. An average of 15.6 mg/kg sodium valproate (range, 10.0-25.5 mg/kg) per day was administered. In some cases the EEG discharge activity continued to be lowered even after the drug blood level had reached steady state. With medication, frequency and total duration of paroxysms were significantly lowered (by more than 90%) in over 80% of the patients, whereas the mean duration of paroxysmal activity did not change uniformly. Before treatment, short paroxysms (1-5 s) were seen together with longer ones in 11 patients. During treatment either all paroxysms disappeared or, in cases of remaining activity, most discharges were short and not accompanied by seizure manifestations. Blood level profiles over 24 h showed maximal values between midnight and 2 a.m. The minimal values (about half of the maximum) were found between 10 p.m. and midnight. The 8 a.m. value was 70-80% of the maximum. Only three patients complained of slight side effects (temporary drowsiness, loss of hair). Because of the simplified handling, the relatively low dose per day, and the few side effects, it seems possible that for primary generalized epilepsies once daily evening administration of sodium valproate is appropriate without diminishing the antiepileptic effect.