Outcomes of seven treatment trials comparing cognitive behavioral therapy to treatment with tricyclic antidepressant medication in major depressive disorder have been quite similar to one another. This led us to question whether treatment outcome in time-limited studies reflected a unique effect of cognitive behavioral therapy. To test the uniqueness hypothesis, relaxation training, a nonpharmacologic, noncognitive treatment, was chosen as a comparison for cognitive behavioral therapy as well as drug therapy. Treatment duration was 16 weeks. The sample of 37 patients treated for major depressive disorder was less depressed than those previously studied. For both cognitive behavioral therapy and relaxation training, outcome of depression was superior to that of tricyclic antidepressant medication by endpoint analysis. The posttreatment scores on the Beck Depression Inventory of 82% of the group receiving cognitive behavioral therapy improved to a Beck Depression Inventory score < or = 9 which was not significantly greater than that for the group receiving relaxation training (73%), so a unique effect was not demonstrated for cognitive behavioral therapy. The outcome for tricyclic antidepressant medication (29% improved to criteria) was significantly worse than that for cognitive behavioral therapy. The patient's pretreatment initial expectancy was not predictive.