The efficacy and safety of midazolam compared with oxazepam and placebo were investigated in 50 hospital patients (19 males, 31 females; age range 21 to 74 years) in a double-blind parallel group study. On the first 2 nights (selection phase), patients received only placebo. On the next 5 nights, they received either 15 mg midazolam, 15 mg oxazepam or placebo. They received no medication on the last 2 nights and were kept under observation (withdrawal phase). Compared with placebo, both benzodiazepines shortened sleep onset latency, reduced the number of awakenings and improved sleep quality. All 3 compounds were well tolerated with only few, mild side-effects (headache, nausea) in the 2 verum groups. Psychometric performance was not impaired on the morning following drug administration. The overall patients' assessments showed 80% satisfaction with midazolam, 66% with oxazepam and 10% with placebo. Midazolam and oxazepam yielded similar results, although midazolam induced sleep more rapidly and was rated more favourably by the patients. Midazolam, in a dose of 15 mg, is thus an effective, fast-acting, well-tolerated hypnotic without residual effects and is suitable for the treatment of insomnia of mild to moderate degree. Oxazepam in a dose of 15 mg is also well suited for the treatment of sleep disorders, particularly if a rapid onset of action is not required.