BACKGROUND: Depression assessment in population studies is usually based on depressive symptoms scales. However, the use of scales could lead to the choice of an arbitrary cut-off point depending on the sample characteristics and on the patient diagnosis. Thus, the use of a medical diagnosis of depression could be a more appropriate approach. OBJECTIVE: To validate a self-reported physician diagnosis of depression using the Structured Clinical Interview for DSM-IV (SCID-I) as Gold Standard and to assess the factors associated to a valid self-reported diagnosis. METHODS: The SUN Project is a cohort study based on university graduates followed-up through postal questionnaires. The response to the question included in the questionnaire: Have you ever been diagnosed of depression by a physician? was compared to that obtained through the SCID-I applied by a psychiatrist or a clinical psychologist. The percentages of confirmed depression and non-depression were assessed for the overall sample and according to several characteristics. Logistic regression models were fitted to ascertain the association between different factors and a correct classification regarding depression status. RESULTS: The percentage of confirmed depression was 74.2%; 95% confidence interval (95% CI) = 63.3-85.1. Out of 42 participants who did not report a depression diagnosis in the questionnaire, 34 were free of the disease (%confirmed non-depression = 81.1%; 95% CI = 69.1-92.9). The probability of being a true positive was higher among ex-smokers and non-smokers and among those overweight or obese but the differences were not statistically significant. CONCLUSION: The validity of a self-reported diagnosis of depression in the SUN cohort is adequate. Thus, this question about depression diagnosis could be used in further investigations regarding this disease in this graduate cohort study.