The aim of the present study was to compare the recently developed polymerase chain reaction (PCR) technique with conventional dot-blot DNA hybridization for human papillomavirus (HPV) detection. Cells were collected by cervicovaginal lavage from a study group of 109 women attending a sexually transmitted diseases clinic. Using a machine that we developed for alternation of temperature cycles, HPV was detected in 51% of patients by PCR. By dot-blot hybridization, 44% of the patients were positive. Concordance of combined positive and negative results between PCR and dot blot was 69%. The greater sensitivity of PCR may have accounted for 19% of specimens that were PCR positive but dot-blot negative. Unexpectedly, however, 12% of specimens were dot-blot positive but negative by PCR, and several specimens were discordant for type of HPV. Both HPV DNA tests agreed with cytology in 41% of women, and in 33% cytology was negative in the face of positive PCR and dot blot. Concordance of cytology with just PCR was 59%, and only with dot blot was 56%. Cervicography agreed with both HPV DNA tests in 41% of patients, with PCR alone in 55%, and with dot blot alone in 58%. Biopsy results did not reveal a strong correlation between histopathological criteria of HPV infection and detection of HPV DNA by either PCR or dot-blot hybridization. Thus the present study has shown that PCR is a slightly more sensitive indicator of HPV infection than dot-blot hybridization. Agreement of HPV DNA results with conventional screening tests was not strong, an observation consistent with many comparative studies by others. In conclusion, PCR is slightly more sensitive than DNA hybridization for detection of HPV, it can be used in conjunction with specimen collection by gentle lavage of the cervicovaginal epithelium, and the possibility remains that it may prove suitable as a screening test.