We studied sleep and daytime function in insomniac patients who took either flurazepam, 30 mg, triazolam, 0.5 mg, or placebo 30 minutes before bedtime. Subjects were 21 patients with either a primary or a secondary diagnosis of chronic psychophysiological insomnia or insomnia associated with personality disorder. Seven subjects were randomly assigned to each condition. The study used a three group by 9 week, double-blind design with three nocturnal sleep recordings each week. During week 1, subjects took no capsules; week 2, subjects took placebo; weeks 3 to 7, flurazepam, triazolam, or placebo; weeks 8 and 9, placebo. Daytime tests for alertness and performance were administered during weeks 1, 3, 5, 7, and 8. Flurazepam showed hypnotic efficacy for weeks 3 to 5. Triazolam showed hypnotic activity for weeks 3 to 7. Although not significant overall, discontinuation of flurazepam produced rebound insomnia in six of seven subjects sometime during the two withdrawal weeks. The relationship between plasma concentration of desalkylflurazepam, the principal active metabolite of flurazepam, and sleep disturbance suggested that the onset of the rebound insomnia depended on the rate of drug washout. Discontinuation of triazzolam produced a significant rebound insomnia on the first and second nights of drug withdrawal. Placebo subjects showed improved sleep throughout weeks 2 to 9 of the study. Daytime testing revealed significantly decreased daytime alertness and decreased performance for flurazepam subjects during weeks 3 to 7, although these effects reverted toward baseline despite continued drug administration.