Culture-negative infective endocarditis (CNE) is a diagnostic problem in spite of improved echocardiographic and blood culturing techniques. We conducted the present study to estimate the proportion of CNE in patients with infective endocarditis, to investigate data regarding risk factors, and to evaluate the Duke and the modified Beth Israel criteria in patients with CNE. We evaluated 820 consecutive suspected episodes of infective endocarditis in adults at the Departments of Infectious Diseases in Göteborg and Borås, Sweden (1984-1996). All patients were diagnosed and treated according to a protocol; 487 episodes were identified as infective endocarditis. Episodes with absence of bacterial growth at blood culture were defined as CNE and were classified with the Duke and the modified Beth Israel criteria. We identified 116 CNE episodes (median age, 67 yr). Mortality was 7%, and in 15%, cardiac surgery was performed. The Duke criteria classified 20 definite, 80 possible, and 16 reject episodes. The modified Beth Israel criteria distinguished 13 definite, 15 probable, 27 possible, and 61 reject episodes. The proportion of CNE among patients with infective endocarditis varied from 19% to 27% at the 2 departments. Antibiotic treatment preceded blood culture in 45% of the CNE episodes. About 20% in a Scandinavian population of infective endocarditis patients have CNE. Antibiotic pretreatment explains less than 50% of all CNE episodes. The Duke criteria are more sensitive but less specific than the modified Beth Israel criteria in classifying patients with CNE.