Sixty-three outpatients with chronic insomnia were treated for 3 weeks under double-blind conditions with either brotizolam (n = 29) at a dose of 0.25 mg or 0.5 mg or placebo (n = 34). A 3-day placebo period preceded and followed the double-blind treatment phase. Brotizolam consistently produced significantly more sleep improvement than placebo but also more adverse effects. In those patients switched abruptly from brotizolam to placebo, rebound insomnia was observed, being most marked at the first post-brotizolam placebo night.