The aim was to evaluate "16S rDNA PCR and sequencing" (PCR) for identification of bacterial DNA in heart valves in routine diagnosis of infective endocarditis (IE). Heart valves from 74 patients with suspected infective endocarditis, and 16 controls were analysed by histology, culture and PCR. Results from blood culture served as the gold standard. Patients were classified according to the Duke criteria. The final classification resulted in 57 definitive cases of IE, 7 possible, and 10 cases without IE. Sensitivity of valve culture was 26% and specificity 62%. Sensitivity of PCR was 72% and specificity 100%. In patients who had received antibiotic treatment for less than 5 days before surgery, sensitivity of culture and PCR were comparable. In patients who had received antibiotic treatment for more than 5 days, sensitivity of valve culture was markedly reduced compared to sensitivity of PCR. In three of seven blood-culture-negative cases PCR was positive, including two cases with non-cultivable bacteria. No PCR samples were contaminated, whereas 35% of valve-culture samples were contaminated. PCR is more sensitive and specific than valve culture, and a valuable supplement to the existing analyses of valve tissue. PCR is necessary to identify the full spectrum of pathogens causing IE. In contrast to sensitivity of culture, sensitivity of PCR was independent of length of antibiotic treatment before surgery.