A prospective study of the epidemiology of infective endocarditis (IE) in a well-defined urban population of 428,000 inhabitants during a 5-year period was carried out. All patients were treated in the same institution, and history, diagnostic procedures, and treatment were standardized. Of 233 consecutive suspected episodes of IE, 127 fulfilled the modified von Reyn criteria. After patients not living in the defined area were excluded, 99 episodes in 90 patients were analyzed in the epidemiologic part of the study. Of these, 33 episodes were definite endocarditis, verified by surgery or autopsy; 35 probable; and 31 possible endocarditis episodes. Another 34 episodes were found retrospectively and are included in the incidence calculation. The crude incidence was calculated to be 6.2/100,000 inhabitants per year, which is high compared to earlier studies. Adjusted to the population of Sweden, the incidence was 5.9/100,000 inhabitants per year. The annual incidence was higher for women, 6.6/100,000, than for men, 5.8/100,000. In the oldest age-group (80-89 years) the annual incidence was 22/100,000 in the prospective study and 30/100,000 if retrospective cases were included. Contrary to almost all other studies, we did not find a male predominance among our cases. Only 7% of patients were intravenous drug abusers, and 15% had a prosthetic valve. The most common bacteria were methicillin-susceptible Staphylococcus aureus (31%) and alpha-streptococci (28%); 12% of episodes were culture negative. The mortality from IE in the population was 1.4/100,000 inhabitants per year. A higher-than-expected incidence of IE was found, especially among older patients and women.