Lorazepam, 4 mg, was evaluated in an 18-night sleep-laboratory study involving five insomniac subjects. Hypnotic effectiveness and effects on sleep stages and related parameters were assessed. Placebo was given on baseline nights 1 to 4, lorazepam on nights 5 to 11, and placebo was given again on withdrawal nights 12 to 18. Subjective and objective data clearly demonstrated that lorazepam was effective for both inducing and maintaining sleep. Sleep latency was reduced from a baseline value of 34.6 min to 17.9 min (P less than 0.01) and total wake time was reduced from 75.9 to 38.5 min (P less than 0.01). On the third and fifth nights of drug withdrawal total wake time rose above baseline levels (termed rebound insomnia) and sleep latency increased by 77% and 60% over baseline (P less than 0.01). Subjective estimates of daytime anxiety also increased above baseline (rebound anxiety) during the withdrawal period. All subjects experienced severe hangover and varying degrees of impaired functioning during the first 3 days on drug. Three subjects also experienced anterograde amnesia during the day after the first drug night. These side effects diminished in intensity over the course of the study. Our results suggest that while 4 mg lorazepam may be effective in inducing and maintaining sleep, this dose induces clinically significant side effects that are followed by consistent rebound phenomena after withdrawal.