The safety and efficacy of 0.25 mg triazolam were compared to those of 15 mg flurazepam and placebo in 41 geriatric outpatients suffering from insomnia. The patients were randomly assigned to one of the three treatment groups. The medication was taken at bedtime for 28 days. Tolerance development was also studied. Triazolam was found to be significantly better than placebo in how much the medication helped the patients sleep, in sleep onset, in duration of sleep, number of nighttime awakenings, in quality (depth) of sleep, and in feeling of restfulness in the morning. Triazolam was significantly better than flurazepam in duration of sleep and was rated higher than flurazepam in all other variables. Flurazepam was significantly better than placebo in only two variables--onset of sleep and quality (depth) of sleep. Side effects were reported in each treatment group, and one patient on placebo discontinued because of side effects. There was no decrease in hypnotic effect over this four-week period, therefore, no evidence of tolerance development on either of the active compounds. Both active compounds provided the same amount of relief from insomnia after four weeks as they had after one week. Laboratory analyses and poststudy physical examinations showed no deleterious effects over the four-week period.