Systematic screening of death certificates referable to residents of two health districts covered by a psychiatric case register allows the study of excess mortality in unselected psychiatric patients from a defined area of known population. Deaths among the case register patients exceeded those expected by 80%; the relative risk was maximal during the first year after registration and was significantly raised in both sexes and in patients of all age-groups. Patients may be selected by service use and by diagnostic group: the excess mortality of "inpatients" both in relative and absolute terms exceeded that of "outpatients". For dementia patients both the relative risk and more notably the absolute risk was high, and the excess mortality was significantly raised in both sexes and all age-groups. It was again greatest during the first year after registration. Deaths of patients with schizophrenia and paranoid psychosis exceeded those expected by 40 per cent and the excess was limited to the first year after registration and occurred predominantly among men.