We reviewed the records of patients with infectious endocarditis from three hospitals in the same city from 1970 to 1972 and from 1980 to 1982. A total of 43 episodes of infectious endocarditis occurred in 42 patients. The clientele of these hospitals differ in that one is a teaching hospital which treats principally indigent tertiary care patients, one treats principally private patients, and one provides care for military veterans. In our study we noted that changes in the frequency and etiology of endocarditis observed at our teaching hospital were not seen in either of the other hospitals. We also noted that the predilection for men and mitral valve involvement observed in some retrospective reviews of endocarditis from large metropolitan centers could not be extended to our city's hospitals. Although some interhospital variations in epidemiologic and microbiologic parameters in infectious endocarditis may be explained by different patient populations, others, such as local variations in the frequency of valvular involvement, cannot be reliably predicted from published series. We conclude that a larger data base founded on statewide reporting of infectious endocarditis would lead to a better understanding of the microbiology, anatomy, and demography of infectious endocarditis.