BACKGROUND In the elderly the impact of atrial fibrillation on mortality and morbidity is substantial. Oral anticoagulant therapy reduces the risk of stroke by 70%; nevertheless, it remains largely underused. We evaluated, in a community prospective study, the factors associated with embolic events and death and the feasibility of oral anticoagulant therapy managed by general practitioners. METHODS We enrolled at the Trieste Cardiovascular Center 229 patients aged > or = 65 years with non-rheumatic atrial fibrillation. At baseline, each patient underwent a transesophageal echocardiography and received instructions about oral anticoagulation. Patients were regularly followed by their general practitioner and finally evaluated at the Center. RESULTS At baseline, the mean age was 73 years, 14% of patients were free of heart disease, 27% had had a previous embolic event, and 33% had an atrial thrombus. After a 5-year follow-up, 85% of the patients had been admitted to hospital, 17% had suffered an embolic event, and 35% were dead. Diabetes and the presence of a low flow in the left atrial appendage were predictive of embolic events. Heart failure, spontaneous echocontrast and aortic plaques were predictive of death. Anticoagulant therapy increased from 14 to 34% but the incidence of major bleeding did not change. The patients on anticoagulant therapy at follow-up constituted the group with the worst cardiovascular profile and embolic rate, but had a lower death rate (19%) compared with those on antiplatelet therapy (32%) and with those without antithrombotic therapy (67%). The hospitalization rates were respectively 78, 83 and 100%. CONCLUSIONS In a group of elderly patients followed by their general practitioner with the support of a specialized cardiologic unit, oral anticoagulant therapy was well tolerated and associated with a significant decrease in mortality and hospitalization.