Sixty volunteers with insomnia participated in a randomized, double-blind, controlled clinical trial. After an initial six nights of placebo, 30 subjects (the abrupt-withdrawal group) received 0.5 mg of triazolam nightly for 7 to 10 nights, after which they received placebo. The other 30 subjects (the tapered-dosage group) received the same initial placebo treatment, then triazolam at 0.5 mg for seven nights, at 0.25 mg for two nights, and at 0.125 mg for two nights, and then placebo. As compared with the initial placebo period, the triazolam period significantly reduced the interval before the onset of sleep (sleep latency), and it prolonged sleep duration, reduced the number of awakenings, and improved the self-rated soundness of sleep in all cohorts. In the abrupt-withdrawal group, plasma levels of triazolam were undetectable the morning after the first night of placebo substitution, and subjects reported prolongation of sleep latency (57 minutes longer than base line), reduction in sleep duration (1.4 hours less than base line), and increased awakenings (1.2 per night above base line). The symptoms of rebound sleep disorder lasted one or possibly two nights, and there was a reversion toward base line on subsequent placebo nights. In the tapered-dosage group, however, plasma triazolam levels fell gradually to zero, and rebound symptoms were decreased or eliminated. Thus, rebound sleep disorder following abrupt discontinuation of triazolam can be attenuated by a regimen of tapering.