Trials of hypnotic medications typically determine efficacy by examining changes in polysomnographically recorded sleep and in daytime performance. The authors employed daytime sleepiness as a new, potentially crucial criterion in such trials. Oxazepam and flurazepam were effective in improving some polysomnographically defined measures of nocturnal sleep in 14 patients with chronic insomnia; flurazepam produced substantial daytime sleepiness and oxazepam did not. Oxazepam produced some rebound insomnia, consisting of about an hour's reduction of polysomnographically defined sleep, but without gross mood disturbance or the patients' awareness of sleep loss.