As part of a study on the influence of some antiepileptic drugs on the sleep of epileptic patients, we investigated the changes of sleep spindles and spike wave discharges. 8 Patients with generalized epilepsy whose polygraphic sleep records before drug treatment were known, received PB and DPH in consecutive monotherapy. Sleep was investigated with steady state therapeutic levels of both drugs. Spindle activity, as compared with the base-line night, was increased with PB and decreased with DPH (p less than 0.05). Three of the 8 patients had slow spindles (6--9/s), together with diminished sigma spindles (12--14/s). Obvious influences of the drugs on these slow spindles were not observed. A relationship between spindle activity and spike wave discharge was indicated by the fact that those 5 patients in whom spike wave activity was significantly diminished or abolished by PB treatment, had a more than 50% increase of spindles, whereas the 3 other had not. With DPH, however, there was no systematic relation between spindles and spike waves. It is therefore considered that the relationship between both phenomena is not very close or causal, and that the effect of DPH on spike wave discharge might be partly different from the effect of PB.