This study was designed in order to devise fitting conditions in polymerase chain reaction (PCR)-in situ hybridization (ISH) for observing human papillomavirus (HPV) infection morphologically in uterine cervical neoplasias and to compare the detection rates of HPV by PCR-ISH and solution phase PCR (S-PCR) as well as fluorescence ISH (FISH). Tissues were obtained from 23 patients with cervical intraepithelial neoplasia 3, who visited our hospital between 1994 and 1997. To detect HPV-16, a HPVpF forward primer and a HPV p 16 reverse primer were used. Compared with the traditional methods, the PCR-ISH technique performed in this study was contrived as follows. To prevent detachment, the specimens were attached to silane-coated slides at 90°C and successi­vely left at room temperature for 36 hours. In proteopepsis, pepsin was used. PCR products were fixed with 4% paraformaldehyde. PCR-ISH, S-PCR, and FISH showed HPV-16 positivity in 52.2%, 56.5% and 21.7%, respectively. The positive rate of HPV-16 detected by PCR-ISH as well as S-PCR was significantly higher than that by FISH (p < 0.01, respectively). There was no significant difference between the positive rates of HPV-16 detected by PCR-ISH and S-PCR. HPV-16 was detected by S-PCR in all 12 spe­cimens in which HPV-16 expression was judged as positive using PCR-ISH. Similarly, HPV-16 was found by PCR-ISH in all five specimens in which HPV-16 expression was regarded as positive using FISH. While the FISH technique detected HPV-16 signals only in the superficial and middle layers of squamous cells, the PCR-ISH technique demonstrated them in all the layers including the parabasal and basal layers. The PCR-ISH technique contrived in this study has a high sensitivity to HPV-16 equal to that of S­PCR. The difference in detection rate and distribution of HPV DNA between PCR-ISH and FISH might suggest that HPV does not infect the superficial layer but rather the parabasal layer.