Fifty-four pairs of cervical biopsies ranging from minimal dysplasia to severe dysplasia were studied for the presence of human papillomavirus DNA by in situ hybridization. Two assays were performed on each biopsy. A 16 hour hybridization was used in one assay, while a 40 hour hybridization was utilized in the second assay. Increasing the hybridization time to 40 hours did not significantly increase the detection rate of HPV compared to the rate found using the 16 hour hybridization. Also, no difference in the detection rate of HPV was found by using one biopsy of the pair over the other biopsy of the pair. However, the performance of a single in situ assay on only one biopsy from each patient significantly underestimated the true prevalence of HPV. A single assay only detected 21/33 (64%) patients with HPV. Implications of multiple testing of all histologically abnormal biopsies is discussed in relation to prospective follow-up studies determining the usefulness of HPV typing in patient management.