We evaluated 355 subjects who entered one of six double-blind placebo-controlled antidepressant drug trials with respect to the occurrence of antecedent adverse life events and their meaning to the patient. Patients were also assessed with regard to the degree of social support they received for the negative life event. The groups differed as to whether they did or did not meet the criteria for melancholic depression; 43 one-week placebo responders were statistically significantly more likely to believe that adverse life events predisposed them to depressive illness and that such life events precipitated their current depression, compared to 312 one-week placebo non-responders. Of the 312 patients who went on to the double-blind phase in which they were treated with either drug (n = 204) or placebo (n = 108), it was noted that, for both melancholic and non-melancholic patients, responders to drug treatment (but not placebo) had a more favourable ratio of social support received/social support desired than non-responders. Non-melancholic responders to both drug and placebo were statistically significantly more likely to report fewer adverse life events and have a less strong belief that adverse life events predispose one to depressive illness than non-responders. Melancholic patients did not show this trend.