PURPOSE: We aimed to study the occurrence and predictors of medical students' mental health problems that required treatment. SUBJECTS AND METHODS: Medical students from all Norwegian universities (N=421) were surveyed in their first term (T1), and 3 (T2) and 6 (T3) years later. The dependent variable was "Mental health problems in need of treatment". Predictor variables included personality traits, medical school stress and negative life events. RESULTS: The lifetime prevalence of mental health problems was 15% at T1. At T2, of the 31% who reported problems during the first 3years, a majority had not sought help. At T3, 14% reported problems during the preceding year. Adjusted predictors of problems at T2 were previous mental health problems (p