Strict case definitions were applied to 123 clinically diagnosed cases of infective endocarditis. Cases were categorized as definite (19), probable (44), or possible (41) endocarditis or were rejected (19). Compared to other published studies, our patients had an advanced mean age (57), high incidence of underlying valvular disease (66%), short mean duration of symptoms (27 days), and 15% mortality, the lowest reported for a large series. Most cases were caused by viridans streptococci, Staphylococcus aureus, or enterococci; Enterobacteriacae were absent, and negative cultures infrequent (5%). Subgroups included nosocomial endocarditis (13%), usually with underlying valvular disease and invasive procedures; prosthesis endocarditis (12%); and cases requiring cardiac surgery (18%). Deaths were caused by heart failure, neurologic events, or superinfection. Strict definitions are useful in managing suspect cases, and are essential in comparing clinical studies. Early recognition and treatment should be the focus of efforts to reduce mortality from endocarditis.