In this 2-year prospective study, we searched for predictive factors influencing the 2-year outcome of major depressive episodes. Demographic characteristics (age, gender, education, employment), illness-related variables (severity, age at onset, number and duration of previous episodes), personality characteristics (DSM-IV personality disorders, trait anxiety, coping style), life context factors (life events before and during the depressive episode, social support, social adjustment), and biological markers (dexamethasone suppression test, thyroid stimulating hormone levels) of 117 inpatients with major depressive episode were assessed. A structural equation model was used to test the proposed correlational structure of the relevant variables. The non-remission of the depressive symptoms by the end of a 6-week acute treatment phase was found to be the most relevant factor predicting sustained non-remission at the end of a 2-year follow-up period. At the end of the sixth week, the severity of depression depended on the level of social support and on the severity of depression at baseline. Among the baseline variables, anxious personality traits and a lower level of education predicted a high level of depressive symptoms at the end of the 2-year follow-up. Life events before and during the depressive episode, and the biological markers at baseline had no direct effect on the outcome. The rapid remission of the depressive symptoms is the most important predictor for the favorable long-term outcome of a depressive episode. Personality characteristics, social support and level of education,--interacting with each other--also play a significant role.