A multicenter, double-blind, sleep laboratory and performance study was conducted to evaluate the hypnotic efficacy and residual effects of midazolam (15 mg) and temazepam (30 mg) compared to placebo when administered in the middle of the night. Eighteen volunteers with objectively verified sleep maintenance insomnia received placebo for 3 nights during week 1 (adaptation and screening). During weeks 2, 3, and 4 they received 2 consecutive nights of midazolam, temazepam, and placebo (one treatment per week) in a balanced crossover design. Treatment was administered in the middle of the night (3.5 hours after bedtime). Neither drug reduced the latency to return to sleep after the middle of the night awakening. Both drugs significantly increased total sleep time, reduced wake during sleep, and number of awakenings over 4.5 hours in bed after treatment. In the morning (5 to 6.5 hours postdrug) significant performance decrements and reduced daytime sleep latency (7 hours postdrug) were found with temazepam but not midazolam.