OBJECTIVE To estimate the incidence of and identify risk factors for hemorrhage and thromboembolism during long-term anticoagulant therapy. METHODS We conducted a population-based retrospective cohort study of all residents of Rochester, Minnesota, in whom a course of warfarin therapy intended to last for more than 4 weeks was initiated between Sept. 1, 1987, and Dec. 31, 1989. METHODS Medical records were reviewed, and pertinent data were compiled. All bleeding complications were classified as minor or major on the basis of the bleeding severity index, and thromboembolic events were classified as major if they were fatal or life-threatening. Cumulative incidences of adverse events were analyzed statistically. RESULTS During the study period, 261 patients had incident courses of anticoagulation (52% were male, 61% were 65 years of age or older, and 31% were 75 years of age or older), with 221 patient-years of warfarin exposure. The primary indications for anticoagulation were venous thromboembolism (39%); stroke or transient ischemic attack (21%); atrial fibrillation (11%); and coronary artery disease, procedures for coronary artery disease, or cardiomyopathy (7%). The cumulative incidence of major hemorrhage at 1, 3, 12, and 24 months was 1.6%, 3.3%, 5.3%, and 10.6%, respectively, and of major or minor thromboembolic events was 2.3%, 5.0%, 7.4%, and 13.1%, respectively. In multivariate analysis, (1) a malignant condition was significantly associated with major hemorrhage; (2) malignant disease and history of peptic ulcer were significantly associated with the combined outcome of major or minor hemorrhage; and (3) malignant disease was significantly associated with any thromboembolism. Age, sex, atrial fibrillation, history of gastrointestinal hemorrhage, history of peptic ulcer, alcohol abuse, hypertension, stroke, and the Charlson comorbidity index were not significantly associated with major hemorrhage. CONCLUSIONS In this population-based study, including a high proportion of elderly patients, malignant disease at initiation of warfarin anticoagulation was significantly associated with both major hemorrhage and any thromboembolism. Advanced age is not a contraindication to anticoagulant therapy.