This study was designed to develop improved criteria for the diagnosis of infective endocarditis and to compare these criteria with currently accepted criteria in a large series of cases. A total of 405 consecutive cases of suspected infective endocarditis in 353 patients evaluated in a tertiary care hospital from 1985 to 1992 were analyzed using new diagnostic criteria for endocarditis. We defined two "major criteria" (typical blood culture and positive echocardiogram) and six "minor criteria" (predisposition, fever, vascular phenomena, immunologic phenomena, suggestive echocardiogram, and suggestive microbiologic findings). We also defined three diagnostic categories: (1) "definite" by pathologic or clinical criteria, (2) "possible," and (3) "rejected." Each suspected case of endocarditis was classified using both old and new criteria. Sixty-nine pathologically proven cases were reclassified after exclusion of the surgical or autopsy findings, enabling comparison of clinical diagnostic criteria in proven cases. Fifty-five (80%) of the 69 pathologically confirmed cases were classified as clinically definite endocarditis. The older criteria classified only 35 (51%) of the 69 pathologically confirmed cases into the analogous probable category (p < 0.0001). Twelve (17%) pathologically confirmed cases were rejected by older clinical criteria, but none were rejected by the new criteria. Seventy-one (21%) of the remaining 336 cases that were not proven pathologically were probable by older criteria, whereas the new criteria almost doubled the number of definite cases, to 135 (40%, p < 0.01). Of the 150 cases rejected by older criteria, 11 were definite, 87 were possible, and 52 were rejected by the new criteria. Application of the proposed new criteria increases the number of definite diagnoses. This should be useful for more accurate diagnosis and classification of patients with suspected endocarditis and provide better entry criteria for epidemiologic studies and clinical trials.