||Objectives: To estimate mortality of patients with affective disorder commenced on lithium and followed up for two to 23 years. To estimate the effects of prolonged exposure to lithium and of regular attendance to a specialized facility. Design: Naturalistic prospective study of a series of consecutively admitted outpatients. Setting: Outpatient unit for the management of lithium and related treatments. Participants: All 1411 patients admitted between January 1, 1980 and June 30, 2000 were studied. Vital status on December 31, 2002 or date of death were established for 1394 (99%) participants. Follow-up totaled 18154 patient-years, including 5474 years of controlled lithium treatment (serum concentrations within 0.5-1.0 mmol/L and intervals between visits no longer than four months). Main outcome measures: The year-by-year mortality of all participants was compared with that of the general Sardinian population, standardized for age, gender and calendar year. Standardized mortality was also analyzed by length of lithium exposure and by regularity of attendance. Results: Two hundred participants died during the observation period. Overall mortality was twice that of the general population. Controlled treatment with lithium for more than five years reduced mortality to the rates expected for the general population. Death rates doubled again in participants who had dropped out after years of initial regular attendance. Conclusions: Lithium can protect patients with affective disorders from otherwise high mortality, provided that adherence to prolonged prophylaxis is controlled and encouraged at a specialized facility. Further research of any protective effect by currently available alternatives to lithium is suggested.