Triazolam, 0.4 and 0.8 mg, flurazepam, 15 and 30 mg, and placebo were compared in a double-blind, randomized 5-night crossover study in 25 inpatient insomniacs. These patients all complained difficulty falling asleep; all said they usually slept less than 5 hr a nigh and woke up too early in the morning. Results of the patients' global evaluation of the medications shows that all of the treatments were rated significantly higher than placebo, with the exception of triazolam, 0.4 mg, which was not significantly different from flurazepam, 15 or 30 mg, or from placebo. In subjective evaluation of sleep onset, only triazolam, 0.4 and 0.8 mg, was rated faster than placebo. All 4 active medications increased duration of sleep. Triazolam, 0.8 mg, and flurazepam, 30 mg, were rated as providing deeper sleep than placebo while all treatments except flurazepam, 15 mg, decreased the number of awakenings below that on placebo. A significant dose-response curve was obtained with triazolam and flurazepam for some of the parameters. Very few adverse effects were reported. One patient reported feeling groggy and drowsy on 0.4 mg triazolam while 2 reported nightmares on placebo.