The etiologic association of human papillomavirus (HPV) with uterine cervical cancer has prompted the need for improved laboratory diagnosis of this virus. The application of conventional hybridization technology, including filter in situ hybridization (FISH) and Southern-blot analysis, has revealed that the detection and typing of the virus is inconsistent between sequential specimens from the same individual. To determine whether the polymerase chain reaction (PCR) can be used to provide a more accurate assessment of infection status, two exfoliated cervical cell specimens obtained sequentially from a cohort of 30 women without clinical evidence of cervical abnormalities were analyzed in parallel by FISH and PCR at 6-month intervals. Neither of the procedures provided consistent findings with two sequential specimens suggesting that multiple analyses are necessary to assess infection accurately. However, PCR was less subjective in interpretation and demonstrated greater specificity than did FISH. With the increased sensitivity inherent to PCR, our findings indicated that PCR is more likely to identify latent HPV infection with a single specimen.