The efficacy of fluvoxamine was compared to that of amitriptyline in a double-blind 6-week fixed-dose trial of 56 inpatients with major depressive episode. The two drugs were comparable in their antidepressant efficacy. We tested the percentage of improvement in Hamilton-D scores during the first and the second weeks of treatment as predictors of efficacy for the last week. Improvement rates during the second week significantly predicted the outcome. We also investigated whether or not some symptomatological characteristics would permit prior prediction of the outcome with amitriptyline or fluvoxamine, dividing our sample into responders and nonresponders to the two drugs. The four groups showed differences in their symptomatological profiles.