The survival of aged Finns suffering from dysthymic disorder (DSM-III criteria) was assessed in two cohorts (60+ yrs and 65+ yrs) in a longitudinal epidemiological study conducted in Ahtäri in western Central Finland from 1984 onwards. The mortality of dysthymic persons (N = 214 and N = 115) was compared to that of the non-depressed population living in the same municipality (N = 982 and N = 853). Two follow-up periods (6 yrs and 11.5 yrs) were used for the first cohort, and one (6 yrs) for the second cohort. The Kaplan-Meier procedure and Mantel-Cox statistics followed by Cox proportional hazards models were used in the analyses. The occurrence of dysthymic disorder in men was related to higher mortality in both cohorts during all follow-up periods. In women, the occurrence of dysthymic disorder was related to higher mortality in the first cohort during the follow-up of 11.5 years, and in the second during the follow-up of 6 years. When age, sex, marital status, education, smoking, physical health and functional abilities were taken into account in the Cox proportional hazards models, high age, male sex, smoking, low educational level, the use of more than two medicines and lowered functional abilities emerged as predictors of mortality in the first cohort during both follow-up periods. In the second cohort, high age male sex, smoking, poor physical health and lowered functional abilities emerged as predictors. The results suggest that the higher mortality of the aged suffering from dysthymic disorder is explained by the high occurrence of somatic diseases and disabilities in dysthymic persons. They do not suggest that there might exist biochemical factors in the aetiology of dysthymic disorders that would increase mortality. Nor do they give any evidence to suggest that dysthymic disorders might be precursors of somatic diseases increasing mortality.