objective: To review (1) the clinical epidemiology of bleeding during anticoagulant therapy with heparin or warfarin, (2) data useful in estimating the risk for bleeding in individual patients, and (3) the efficacy of methods for its prevention. methods: Relevant literature was identified by a computerized search of the Medline database and by review of the bibliographies of original and review articles. Studies were classified according to their design. Estimates of the risk for bleeding during anticoagulant therapy, compared with the risk without therapy, were obtained from randomized trials. Estimates of the frequency of bleeding during the course of anticoagulant therapy and information about risk factors for bleeding were obtained primarily from longitudinal studies of inception cohorts of patients followed from the start of therapy. main results: The average daily frequencies of fatal, major, and major or minor bleeding during heparin therapy were 0.05%, 0.8%, and 2.0%, respectively; these frequencies are approximately twice those expected without heparin therapy. The average annual frequencies of fatal, major, and major or minor bleeding during warfarin therapy were 0.6%, 3.0%, and 9.6%, respectively; these frequencies are approximately five times those expected without warfarin therapy. The risk for anticoagulant-related bleeding is highest at the start of therapy: during warfarin therapy, the risk for major bleeding during the first month of therapy is approximately 10 times the risk after the first year of therapy. An individual patient's risk for major anticoagulant-related bleeding can be estimated on the basis of specific risk factors such as the intensity of the anticoagulant effect achieved and the presence of serious comorbid diseases, especially cerebrovascular, kidney, heart, and liver disease; older age and concurrent medicines may also be independent risk factors. Major bleeding most often affects the gastrointestinal tract, soft tissues, and urinary tract. Diagnostic evaluation of gastrointestinal bleeding and gross hematuria leads to identification of previously unknown lesions in approximately one-third of cases, even when the prothrombin time is elevated. Intracranial bleeding is rare, but it is frequently fatal. The frequency of bleeding during warfarin therapy is reduced by less intense therapy achieving a prothrombin time with an International Normalized Ratio of 2.0 to 3.0, which is efficacious for most indications. conclusion: Anticoagulant-related bleeding is common and often serious. The risk for bleeding can be estimated in an individual patient, giving the primary physician a quantitative basis for weighing the risks and benefits of therapy and for optimizing patient management. The frequency of anticoagulant-related bleeding is reduced by less intense warfarin therapy. Future studies should evaluate new approaches to management that may further reduce complications while maintaining efficacy. © 1993.