AbstractDiazepam is one of the most widely used, broad‐spectrum anti‐anxiety agents. The aim of this study was to evaluate the efficacy of diazepam, and to establish whether it is more effective than a placebo in improving the various neurotic anxiety states seen in patients with neurosis or psychosomatic disease. Of the recently established comprehensive register of psychotropic drug trials carried out in Japan, a total of 17 double‐blind, randomized controlled trials were identified on the treatment of neurosis using anti‐anxiety compounds, in which both diazepam and placebos were used. Meta‐analysis of these 17 studies demonstrated that diazepam is significantly more effective than a placebo (relative risk 1.35, 95% confidence interval 1.21–1.51, number needed to treat 9). The maximal effective dose of diazepam seems to be 12 or 18 mg/day with a treatment duration of 2 or more weeks. There was no significant difference between the effects of placebo and a diazepam dose of 6 mg/day. Caution should be exercised in assessing these results, however, since this is the first meta‐analysis showing the significant effectiveness of diazepam in the treatment of neurosis or psychosomatic disease. Copyright © 2003 John Wiley & Sons, Ltd.