Objective. - To update recommendations issued by the American Heart Association last published in 1990 for the prevention of bacterial endocarditis in individuals at risk for this disease. Participants. - An ad hoc writing group appointed by the American Heart Association for their expertise in endocarditis and treatment with liaison members representing the American Dental Association, the Infectious Diseases Society of America, the American Academy of Pediatrics, and the American Society for Gastrointestinal Endoscopy. Evidence. - The recommendations in this article reflect analyses of relevant literature regarding procedure-related endocarditis, in vitro susceptibility data of pathogens causing endocarditis, results of prophylactic studies in animal models of endocarditis, and retrospective analyses of human endocarditis cases in terms of antibiotic prophylaxis usage patterns and apparent prophylaxis failures. MEDLINE database searches from 1936 through 1996 were done using the root words endocarditis, bacteremia, and antibiotic prophylaxis. Recommendations in this document fall into evidence level III of the US Preventive Services Task Force categories of evidence. Consensus Process. - The recommendations were formulated by the writing group after specific therapeutic regimens were discussed. The consensus statement was subsequently reviewed by outside experts not affiliated with the writing group and by the Science Advisory and Coordinating Committee of the American Heart Association. These guidelines are meant to aid practitioners but are not intended as the standard of care or as a substitute for clinical judgment. Conclusions. - Major changes in the updated recommendations include the following: (1) emphasis that most cases of endocarditis are not attributable to an invasive procedure; (2) cardiac conditions are stratified into high-, moderate-, and negligible-risk categories based on potential outcome if endocarditis develops; (3) procedures that may cause bacteremia and for which prophylaxis is recommended are more clearly specified; (4) an algorithm was developed to more clearly define when prophylaxis is recommended for patients with mitral valve prolapse; (5) for oral or dental procedures the initial amoxicillin dose is reduced to 2 g, a follow-up antibiotic dose is no longer recommended, erythromycin is no longer recommended for penicillin-allergic individuals, but clindamycin and other alternatives are offered; and (6) for gastrointestinal or genitourinary procedures, the prophylactic regimens have been simplified. These changes were instituted to more clearly define when prophylaxis is or is not recommended, improve practitioner and patient compliance, reduce cost and potential gastrointestinal adverse effects, and approach more uniform worldwide recommendations.