PCR has proved superior to conventional blood culture for diagnosing bacteraemia in the presence of antibiotics. Nevertheless, even PCR might yield false-negative results if the template DNA were to be cleaved by serum DNAases after antibiotics had induced bacterial death. To evaluate the cleavage of bacterial template DNA by human serum DNAase I, serum samples inoculated with purified Escherichia coli DNA were incubated with increasing amounts of recombinant human DNAase (rhDNAase) and then examined by a PCR specific for E. coli. As a prerequisite of potential DNAase attack, the release of E. coli DNA after antibiotic-induced bacterial death was quantified by fluorescence microscopy and flow cytometry. Finally, the influence of rhDNAase on the PCR-based detection of antibiotic-killed E. coli in serum was assessed. The results indicated that purified E. coli DNA is remarkably stable in human serum; positive PCR results did not decrease significantly until the ratio of recombinant human DNAase I:E. coli rose to 106:1. As only 14.8-28.4% of the total E. coli DNA was released after antibiotic killing, the PCR-based detection of E. coli fell by only 10% when cefotaxime-killed E. coli were incubated with rhDNAase. It was concluded that human serum DNAases and antibiotic killing do not compromise the reliability of PCR examinations for bacteraemia.