The objective of this study was to determine the validity of human papillomavirus (HPV) detection using exfoliated cervical cells compared with cervical biopsy specimens in women with normal cervix and to assess whether HPV detection rates using exfoliated cells is dependent on the number and order in which cervical scrapes are taken. Women undergoing hysterectomy for reasons other than cervical cancer were recruited in three hospitals in countries with varying risks of cervical cancer. After informed consent and at the time of surgery, three consecutive cervical scrapes were taken as well as four biopsy specimens, one in each of the quadrants around the cervical os. In this study, 331 women were recruited and provided 992 cell samples and 1324 biopsy samples. All scrapes and a sample of biopsy specimens (n = 103) were tested by polymerase chain reaction enzyme immunoassay using a general primer (GP5+/ bio6+). Type-specific tests were performed for 14 HPV types at the subpicogram level in one test and individually. Positive samples were verified using Southern blot hybridization. The prevalence of HPV DNA was 6.3% in cervical cells. Of 19 HPV positive samples in the scrapes, 17 were confirmed in the biopsy specimens. The agreement, as measured by the Kappa statistic, was 0.90 (P < 0.0001). The concordance in detecting HPV infection between scrapes and biopsy specimens was 97.5%, and the concordance in categorizing the samples as negatives was 94.4%. These values were unchanged when the order in which scrapes were taken was compared. Among women without cervical cancer, HPV DNA detection rates do not vary if exfoliated cells or random biopsy specimens are taken as the primary testing specimen. Screening programs based on highly sensitive HPV DNA detection technology in cell scrapes should expect a minimal underdetection.