Background. Few community-based studies have examined the impact of life events, life conditions and life changes on the course of depression. This paper oxamines associations of life events on depressive symptom onset, improvement, and stability. Methods. Direct interview data from the Early Developmental Stages of Psychopathology Study (EDSP), a 4-5 year prospective-longitudinal design based on a representative community sample of adolescents and young adults, aged 14-24 years at baseline, are used. Life events were measured using the Munich Event-Questionnaire (MEL) consisting of 83 explicit items from various social role areas and subscales for the assessment of life event clusters categorized according to dimensions such as positive and negative and controllable and uncontrollable. Depressive disorders were assessed with the DSM-IV version of the Munich Composite Diagnostic Interview (M-CIDI). Multiple logistic regression analyses examined the effects of 22 predictors on the course of depression (onset, improvement, stability). Results. Younger age, low social class, negative and stressful life events linked to the family were associated with increased risk of new onset of depression. Anxiety was a significant independent predictor of new onset of depression. Absence of stressful school and family events was related to improvement in depression. The weighted total number of life events predicted stable depression. Conclusions. The association between life events and the course of depression appears to vary according to the outcome being examined, with different clusters of life events differentially predicting onset, improvement, and. stability.