1. Intent-to-treat analysis of a parallel, double-blind comparative study of fluvoxamine, imipramine and placebo in patients with major depressive episode (DSM III). 2. A posteriori the patients have been stratified in three severity groups on basis of their initial HAMD-score: Mild depression (15-20), Moderate depression (21-25) and Severe depression (> or = 26). 3. At endpoint both the decrease in HAMD-scores and the percentage responders (CGI 1-2) differed significantly comparing fluvoxamine with placebo and tended to differ significantly comparing with imipramine in the severely depressed patients. 4. The results with fluvoxamine in severely depressed patients were better than in moderately depressed patients, which in turn were better than in mildly depressed patients. 5. Despite a lower daily fluvoxamine dosage in the mild group, the premature terminator rate in this group (47%) was higher than in the moderate group (27%) or in the severe group (24%). With placebo the situation was opposite. More patients (47%) terminated before day 28 in the severe group, compared to the moderate (32%) and the mild group (24%), while on placebo. 6. The majority of responders to fluvoxamine at endpoint, showed already improvement at week 1 (82.5%).