Two hundred forty-one outpatients with a DSM-III diagnosis of major depressive disorder participated in a six-week double-blind therapeutic trial of alprazolam, diazepam, imipramine hydrochloride, and placebo. Side effects were given as a major reason for attrition by patients taking the three active compounds and ineffectiveness was the reason given by patients taking placebo. Imipramine-treated patients reported the most and placebo patients the least number of adverse effects. Imipramine and alprazolam, but not diazepam, produced significantly more improvement in depressed symptomatology than did placebo. Mean diazepam scores frequently assumed an intermediate position between those of imipramine or alprazolam and placebo. These treatment differences were found to be independent of initial severity levels of anxiety and depression.