1 Nine volunteer poor sleepers, of mean age 61 years, took trazodone 150 mg nightly for 3 weeks, preceded by 2 weeks and followed by 1 week of matching blanks, in order to examine the effects of electrophysiologically-recorded and subjectively-rated sleep. The second of the initial weeks of matching blanks served as a baseline week. 2 In the subjective ratings, sleep improved in quality on trazodone, significantly so in the first and second weeks of intake, though with significant rebound insomnia on the second withdrawal night. 3 Trazodone halved the frequency of arousals interrupting sleep, and it reduced the time spent in stage 1 (drowsiness). It increased the duration of slow-wave sleep (stages 3 + 4), with a negative rebound following withdrawal. It reduced the time spent in REM sleep, with a rebound above baseline levels after withdrawal. 4 Trazodone did not change total sleep duration, nor the time required to fall asleep. 5 The effects of trazodone were sustained or became enhanced during the period of intake. They persisted for over 24 h after the last dose, and rebound effects were maximal on the second withdrawal night.