Midazolam, an investigational hypnotic, was evaluated for effectiveness, side effects, and withdrawal phenomena in doses of 10, 20, and 30 mg in three separate sleep laboratory studies, each including 4 placebo-baseline nights, 7 drug nights, and 3 placebo-withdrawal nights. Only a slight to moderate degree of effectiveness was shown across the three doses; this effectiveness was much more pronounced during the first third of the night. There was no dose-response effect for effectiveness with either initial or continued drug administration. In general, there was less effectiveness on the last 3 drug nights, indicating a potential for the development of tolerance over a relatively short period of time. Following withdrawal there was a marked dose-related worsening of sleep above baseline levels (rebound insomnia).