SynopsisThe object of the study was the assessment of the mortality risk for persons with a mental disorder in an unselected representative community sample assessed longitudinally. Subjects from a rural area in Upper Bavaria (Germany) participated in semi-structured interviews conducted by research physicians in the 1970s (first assessment) and death-certificate diagnoses were obtained after an interval up to 13 years later. The sample consisted of 1668 community residents aged 15 years and over.Cox regression estimates resulted in an odds ratio of 1·35 (confidence interval 1·01 to 1·81) for persons with a mental disorder classified as marked to very severe. The odds ratio increased with increasing severity of mental illness from 1·04 for mild disorders, 1·30 for marked disorders, to 1·64 for severe or very severe disorders. The relative risk (odds ratio) for persons with a mental disorder only and no somatic disorder was 1·22, for persons with only a somatic disorder 2·00, and for those with both a mental and a somatic disorder 2·13. The presence of somatic illness was responsible for most of the excess mortality. Somatic disorders associated with excess mortality in mental disorders were diseases of the nervous system or sensory organs, diseases of the circulatory system, diseases of the gastrointestinal tract, and diseases of the skeleton, muscles and connective tissue (ICD-8).Thus, while mental illness alone had a limited effect on excess mortality, comorbidity with certain somatic disorders had a significant effect.