New diagnostic criteria for infective endocarditis (IE) have been proposed by the Duke University Endocarditis Service (Durham, NC) to update the widely used Beth Israel (Boston) criteria. We compared the Duke criteria with the Beth Israel criteria in a series of 115 consecutive patients with suspected IE who were hospitalized in a referral center. The diagnosis of IE was histologically and/or bacteriologically confirmed for 27 operated patients. If surgery had not been performed on these 27 patients, 22 vs. 12 would have been classified as having ¿clinically definite¿ and ¿probable¿ IE by the Duke vs. the Beth Israel criteria, respectively, whereas 0 vs. 5 would have been ¿rejected¿ by the Duke vs. the Beth Israel criteria, respectively. The improvement in sensitivity of the criteria from 44% (Beth Israel) to 82% (Duke) was statistically significant (P < .01). We confirm that the Duke criteria improve the sensitivity of diagnosis of IE. The specificity of these criteria should be further evaluated.