Sixty patients meeting specific criteria for atypical depression completed six weeks of double-blind, randomly assigned treatment with phenelzine sulfate, imipramine hydrochloride, or placebo. The overall response rates were 67% with phenelzine, 43% with imipramine, and 29% with placebo. At week 6, phenelzine was superior to placebo on many measures, while the superiority of imipramine to placebo was confined to several variables. Phenelzine was superior to imipramine on the interpersonal sensitivity and paranoia factors of the 90-item Hopkins Symptom Checklist, with trends toward superiority on several other measures, while imipramine was not differentially superior on any measure. Atypical depressive patients with a history of spontaneous panic attacks and hysteroid dysphoric patients both showed extremely low rates of response to placebo and high rates of response to phenelzine. Conversely, those without panic or hysteroid dysphoric features responded equally to all three treatments. Responders to pheneizine also had greater platelet monoamine oxidase inhibition while receiving drug therapy than did nonresponders. Completion of the 120-patient sample will allow more detailed analyses.