The prognosis of depression in elderly people is reported as poor, with high mortality and high rate of non-recovery. The aim of this study was to investigate the outcome after 6 years in a broader perspective. In addition to the risk of dying, we also estimated the risk of moving to long-term care, and having a depression in a re-screening event after 6 years. In 1993, a screening event for depressive symptoms using the Hopkins Symptom Checklist (HSCL-25) was carried out on 1215 subjects aged 65 years and older living in a health care district in Karlstad, Sweden. The study population was divided into a high score, a low score and a dropout group. A re-screening of the same population was carried out in 1999. Logistic regression analysis was used for calculating odds ratios for the selected end-points adjusted for age, gender, civil status and occurrence of physical illness. The adjusted odds ratios for dying for the high score group were 2.5 (95% CI 1.5–4.4) and for any of the end-points 6.1 (95% CI 3.5–10.8) compared with the low score group. Nearly three-quarters of the subjects in the high score group were deceased, depressed or had moved to long-term care after 6 years. The prognosis was poor for depressed elderly people despite not having more physical illness. It is believed that the poor outcome is due to an additive effect of depression on the total medical disease burden.