We randomly assigned 425 outpatients, independently classified as primarily depressed by two trained psychiatrists, to double-blind treatment with Imipramine hydrochloride, chlordiazepoxide hydrochloride, or placebo. Those patients who remained at least moderately depressed (following a two-week placebo washout period) were treated for an additional eight weeks. An endpoint analysis of 387 patients who completed two or more weeks of medication disclosed early therapeutic advantages of chlordiazepoxide. By week 4 of treatment, however, imipramine produced more improvement than did placebo and chlordiazepoxide. By six and eight weeks a general, marked therapeutic advantage was found for imipramine relative to placebo and to chlordiazepoxide on measures of depression, anxiety, anger-hostility, interpersonal sensitivity, and global improvement. Chlordiazepoxide-treated patients generally did significantly better on sleep difficulty but significantly worse on anger-hostility and interpersonal sensitivity than did imipramine- or placebo-treated patients.