Human papillomavirus infection is implicated as an etiologic agent in the development of neoplasia and invasive carcinoma of the cervix. To detect human papillomavirus infection of the cervix, cells must be collected and assayed for human papillomavirus-related deoxyribonucleic acid sequences. Gynecologists and other clinical investigators generally use an exocervical spatula scrape and an endocervical swab for cell collection, analogous to Papanicolaou smear collection. However, inadequate cell recovery is common. To overcome this problem, we have developed the cervicovaginal lavage method for human papillomavirus detection. In the present study we compared the cervicovaginal lavage method with the widely used scrape-swab method in 48 women referred for colposcopic examination. After a Papanicolaou test, two samples were obtained from each woman, either with cervicovaginal lavage followed by scrape-swab or with the scrape-swab followed by cervicovaginal lavage. Human papillomavirus types were assessed by restriction analysis and Southern blot hybridization. In 21 women (44%) test results were positive for human papillomavirus with both the scrape-swab and cervicovaginal lavage cell collection methods; in nine women (19%) test results were positive only with the cervicovaginal lavage method; and in 18 women (38%) results were negative for human papillomavirus with both techniques. None of the women had human papillomavirus detected by scrape-swab without also having it detected with cervicovaginal lavage. The human papillomavirus deoxyribonucleic acid types identified were concordant in the 21 women whose infections were detected with both sampling methods, although the second virus type was detected only with cervicovaginal lavage in one woman who had a mixed genital tract infection. We concluded that cervicovaginal lavage is a more sensitive cell collection method than the scrape-swab technique for assessing human papillomavirus infection of the cervix.