OBJECTIVE Psychiatric disorders involve an increased risk of mortality. In Italy psychiatric services are community based, and hospitalization is mostly reserved for patients with acute illness. This study examined mortality risk in a cohort of psychiatric inpatients for 16 years after hospital discharge to assess the association of excess mortality from natural or unnatural causes with clinical and sociodemographic variables and time from first admission. METHODS At the end of 2002 mortality and cause of death were determined for all patients (N=845) who were admitted during 1987 to the eight psychiatric units active in Florence. The mortality risk of psychiatric patients was compared with that of the general population of the region of Tuscany by calculating standardized mortality ratios (SMRs). Poisson multivariate analyses of the observed-to-expected ratio for natural and unnatural deaths were conducted. RESULTS The SMR for the sample of psychiatric patients was threefold higher than that for the general population (SMR=3.0; 95 percent confidence interval [CI]=2.7-3.4). Individuals younger than 45 years were at higher risk (SMR=11.0; 95 percent CI 8.0-14.9). The SMR for deaths from natural causes was 2.6 (95 percent CI=2.3-2.9), and for deaths from unnatural causes it was 13.0 (95 percent CI=10.1-13.6). For deaths from unnatural causes, the mortality excess was primarily limited to the first years after the first admission. For deaths from natural causes, excess mortality was more stable during the follow-up period. CONCLUSIONS Prevention of deaths from unnatural causes among psychiatric patients may require promotion of earlier follow-up after discharge. Improving prevention and treatment of somatic diseases of psychiatric patients is important to reduce excess mortality from natural causes.