PCR and culture were comparatively evaluated for their abilities to demonstrate Francisella tularensis in wound specimens from tularemia patients during an outbreak in Sweden in 1998. For transport of the specimens used for PCR, a buffer solution containing a nuclease inhibitor was used, and for transport of the specimens used for culture, a commercial transport system was selected after experimental comparison of various systems. Of 40 patients with culture- and/or serology-verified ulceroglandular tularemia, PCR detected F. tularensis DNA in 30 (75%) patients, whereas culture detected bacterial growth in 25 (62%) patients. Compared to data from a previous study, the present inclusion of a nuclease inhibitor in the transport medium did not improve the sensitivity of the PCR, whereas the sensitivity of the culture procedure was significantly increased by selection of the system used for transport. Among eight patients with clinically suspected tularemia but with negative serology and culture, specimens from four patients showed detectable DNA. In three of these patients the diagnosis was verified by the demonstration of an F. tularensis-specific T-cell response in vitro. In conclusion, PCR was more sensitive than culture for demonstration of F. tularensis in wound specimens. Besides, we showed that tularemia may proceed without development of serum antibodies, and in these patients, PCR may be of special importance for verification of the diagnosis.