One hundred and sixteen consecutive women attending a Baltimore City STD clinic were studied for the prevalence of human papillomavirus (HPV) infection of the genital tract using three criteria: presence of clinically recognized (visible) genital warts, cytopathologic evidence suggestive of HPV infection in a Papanicolaou smear, and analysis of cervical scrapes for genital tract HPV genomic sequences by Southern hybridization. The women were young (median age: 22 years) and more than 80% had a history of one or more STDs. The prevalences were 17% for visible warts, 41% for cytologic findings suggestive of HPV infection, and 12% for HPV DNA in cervical scrapes. Comparing the results of the three techniques, HPV DNA was found significantly more often in cytopathology-positive women than in cytopathology-negative women (18% vs. 5%, P = 0.05) and in women with visible warts than in women without visible warts (29% vs. 6%, P = 0.01). Visible warts were more common in women with HPV-DNA-positive cervical scrapes than in HPV-negative women (50% vs. 14%, P = .01). Although 52% of women were judged as infected by at least one of the three criteria, only 4% were infected by using all three criteria. The prevalence of infection was 23% if cytopathology alone was excluded as evidence of HPV infection. These results indicate the difficulty in an accurate estimation of the prevalence of HPV infections, even in a high-risk population.