Report on a prospective longitudinal study of alterations in polygraphic sleep by phenytoin monotherapy for epilepsy. A first dose of 100mg already caused abbreviation of sleep latency and an increase of slow-wave sleep in the first NREM-REM cycle. In the course of adjustments to steady state, an increase of Stage 3 + 4 sleep in the later REM cycles developed, such that the percentage of slow-wave sleep for the whole night was also augmented, whereas the percentage of light sleep decreased. Sleep structure was affected particularly in the third NREM-REM cycle. With continuing therapy, however, these effects were reversed. The only permanent effect was an abbreviated sleep latency. There were only minimal differences in the response of generalized and of localization-related epilepsies. Serum drug levels had only a very limited influence, seizure control and length of follow-up had no influence on the results. As a collateral finding, a delayed further decrease of epileptic discharges during sleep was observed under long-term conditions in patients who were seizure-free and had been so since adjustment to the steady state.