Nine subjects who believed themselves to be poor sleepers, of mean age 58 years, took a placebo for 7 days, then ritanserin 5 mg for 20 days, followed by 3 days on placebo. Sleep was recorded electrophysiologically on 2 nights during baseline, 2 early drug nights, 2 late drug nights and the 2nd and 3rd withdrawal nights. Ratings of sleep quality were collected each morning. Ritanserin, a 5HT2 antagonist, caused a persistent doubling of the amount of EEG slow wave sleep, without altering the total duration of sleep. Ritanserin decreased the frequencies of awakening and after about a week it appeared to improve the subjective quality of sleep. Sleep was then impaired during withdrawal, as indicated by decreased duration and poorer subjective quality, being worst on the 3rd withdrawal night.