One hundred thirty-five cases of infective endocarditis in adults at the University of Alberta Hospital from 1985 to 1993 were reviewed and the von Reyn and Duke criteria were compared. There were 80 cases of native valve endocarditis, 15 cases of endocarditis in intravenous drug users, 7 cases of early prosthetic valve endocarditis, and 33 cases of late prosthetic valve endocarditis. Valve replacement or repair was performed in 33% of all cases. The overall mortality was 19%. The mortality among patients treated surgically was significantly lower than that among those treated medically (9% vs. 24%, respectively; P = .037). However, when patients who were too medically unstable for surgery or who refused surgery were excluded, the mortality among the medically treated group decreased to 15%, which was not significantly different from that among the surgically treated group. The 33 patients transferred from other hospitals were infected with similar pathogens; however, the rate of surgical intervention among these patients was much higher than that among other patients (64% vs. 24%, respectively; P < .0001), and the mortality was slightly lower (12.1% vs. 21.6%, respectively; P = .31). The 54 surgical and autopsy-proven cases were classified by the von Reyn and the Duke criteria without knowledge of the operative and autopsy results: 15% of these cases were rejected by the von Reyn criteria, and none were rejected by the Duke criteria.