In this investigation, the antidepressant efficacy of fluvoxamine and imipramine was compared in a randomized, double-blind, placebo-controlled study lasting 4 weeks; 338 depressed patients were recruited at five North American centres. For the efficacy analyses an intent-to-treat sample was defined. The global efficacy of the two drugs was assesssed by the Hamilton Depression scale (HAM-D) and Clinical Global Impression (CGI) scores. Antidepressant activity was also assessed using the percentage of responders on the CGI “improvement” scale. In addition the time of onset of antidepressant effect was evaluated by weekly analysis of individual HAM-D items. The intent-to-treat sample was stratified retrospectively according to the severity of the depression (mild, moderate or severe). Regarding global efficacy, compared with placebo, only fluvoxamine significantly improved the HAM-D total scores at Week 4 (p < 0.05). There was a suggestion from individual HAM-D item scores (depressed mood, suicide, psychic anxiety) that fluvoxamine had an earlier effect than imipramine. Overall, compared with placebo, more HAM-D items were improved by fluvoxamine than imipramine. Fluvoxamine but not imipramine was significantly superior to placebo in severely depressed patients as shown by improvements in the HAM-D score (p < 0.01) and the CGI “improvement” score (p < 0.05). Side effect profiles for the active agents were typical for their pharmacological category: imipramine was associated with anticholinergic effects, particularly dry mouth, and fluvoxamine was associated with nausea and vomiting.