This double-blind, placebo-controlled study compared venlafaxine (immediate release), the first modern serotonin-norepinephrine reuptake inhibitor, with the selective serotonin reuptake inhibitor fluoxetine. Outpatients were randomly assigned to 6 weeks of treatment with venlafaxine (75–225 mg/day; n = 102), fluoxetine (20–60 mg/day; n = 104), or placebo (n = 102). Efficacy was assessed using the 21-item Hamilton Depression Rating Scale (HAM-D21), the Montgomery-Åsberg Depression Rating Scale (MADRS), the Clinical Global Impression-Severity of Illness (CGI-S) scale, response and remission rates, and several other measures. Intent-to-treat analyses utilized both the last observation carried forward and ETRANK methods to account for missing data. At week 6 or study endpoint, venlafaxine (mean dose: 142 mg/day) was superior to placebo on most outcomes measures, whereas the differences between fluoxetine (mean dose: 41 mg/day) and placebo were less consistent. Final remission (defined as HAM-D ⩽ 7) rates were 32%, 28%, and 22% for venlafaxine, fluoxetine, and placebo, respectively. Few differences between the active treatments attained statistical significance. Both active therapies were generally well tolerated; however, attrition due to adverse events, incidence of selected side effects, and increases in pulse and blood pressure favored fluoxetine over venlafaxine. This study provides further evidence that venlafaxine is effective after 6 weeks of treatment compared with placebo. The efficacy profile of fluoxetine was somewhat less consistent. It is strongly recommended that future studies of comparative antidepressant efficacy be adequately powered to detect modest between-drug differences in efficacy.