OBJECTIVE Recurrent episodes of major depressive disorder are reported to have similar or stable characteristics across episodes. However, symptoms appear to be moderately stable only in consecutive depressive episodes or if episode severity is considered. The authors prospectively studied major depressive episodes occurring within 2 years to determine whether symptoms in the second episode could be predicted on the basis of symptoms in the first. METHOD Inpatients (N=78) with major depressive disorder were rated with the Hamilton Depression Rating Scale during two separate episodes. Patients had a baseline assessment at index hospitalization and follow-up visits at 3, 12, and 24 months after discharge. Information regarding the presence of a major depressive episode, its duration, and its severity was documented. Baseline and follow-up data were analyzed by using Pearson correlations with and without adjustments for severity of depression. RESULTS Subtype of depression appeared not to be stable. The most robust, although still weak, correlations across episodes were for anxiety and suicidal behavior. Only modest correlations were identified for a few depressive symptoms. CONCLUSIONS The lack of robust consistency of symptoms or depressive subtype across episodes is striking given the requirement of five of nine predetermined symptoms for depression, increasing the chances of finding an association. These findings suggest that there is a superfamily of mood disorders with pleomorphic manifestations across major depressive episodes within individual patients with unipolar depression.