The purpose of this study is to evaluate the novel antidepressant venlafaxine for the management of treatment-resistant unipolar depression. We gave unblinded venlafaxine to 84 consecutive outpatients and inpatients who met DSM-III-R criteria for major depression and who had failed to respond to at least three adequate trials of antidepressants from at least two different antidepressant classes or electroconvulsive therapy, plus at least one attempt at augmentation. Patients were evaluated after a drug free period at baseline and regular intervals with the 21-item Hamilton Rating Scale for Depression (HAM-D-21), Montgomery-Asberg Depression Rating Scale (MADRS), and the Clinical Global Impressions Scale Improvement item (CGI). Full response for each scale was defined as follows: HAM-D-21 score of 8 or lower, a MADRS score of 12 or lower, and CGI score of 1; partial responses was defined as a 50% decrease in the HAM-D and MADRS, with final scores greater than 8 and 12, respectively, and for the CGI, a score equal to 2. About a third of patients were considered to be either full or partial responders (32.9% by HAM-D-21, 30.0% by MADRS, and 40% by CGI) after 12 weeks of venlafaxine treatment. To date, about 46% of responders have sustained their response for at least 3 months after the acute response. Venlafaxine is effective for a significant, but small, minority of patients with rigorously defined triple-resistant depression; the improvement was maintained for about half of the responders for the first 3 months of maintenance therapy.