Cervical tumours evolve very slowly from precursor lesions known as dysplasias or cervical intraepithelial neoplastic (CIN) lesions, which may regress or persist (for review see Howley, 1991). Most human papillomavirus (HPV) type 16-associated cervical tumours have integrated viral DNA. Several studies have used Southern hybridizations to analyse physical state (Du$ rst et al., 1985 ;Cullen et al., 1991) and RNA in situ hybridization to examine patterns of HPV-16 gene expression. The results suggest that CIN lesions have episomes, and tumours have integrated HPV-16 genomes (for review see zur Hausen, 1991). This is also supported by a study using a PCR assay which reported the presence of episomes in all grades of CIN lesions (Das et al., 1992). This PCR assay is based on detecting disruption of the viral E2 gene, a transcription factor Author for correspondence : Betty Daniel.Fax j91 80 3343851. e-mail betty!ncbs.tifrbng.res.in which is believed to function predominantly as a repressor of the upstream regulatory region (URR) and is found to be disrupted in most cervical tumours (for review see McBride et al., 1991). The implication of detectable episomes in precursor lesions is that integration of the virus correlates with, or is the cause of, progression to invasiveness.However, on the contrary, some groups have previously used Southern hybridization as a technique to analyse physical state and reported the presence of a fair number of integrated genomes in CIN lesions (Di Luca et al., 1986 ; Lehn et al., 1988). Using the PCR assay developed by Das and colleagues, we have also previously reported that we were unable to detect episomes in high-grade CIN lesions (Daniel et al., 1995). In this paper, we have matched the PCR analysis with an RNA in situ hybridization analysis and the results support our previous assertion that the integration event is associated with the development of high-grade CIN lesions. We have also analysed changes in cytokeratins, markers of invasiveness and the Notch signal transduction pathway during the evolution of cervical lesions.The absence of amplification with primers for the E2 gene in the presence of an amplification with primers corresponding to the URR and E6 sequences correlated very well with the presence of integrated HPV-16 DNA (Das et al., 1992). In this study, we undertook a PCR analysis along with RNA in situ hybridization, using standard protocols, with fragments spanning two regions, the first covering the E6-E7 region (65-875) and the second covering the E2-E5-L2 region (3697-4761). In 15 out of 16 CIN III lesions and 19 out of 19 tumours, there were no signals detected with the E2-E5-L2 probe, while signals were detected with the E6-E7 probe in all cases. The PCR analysis showed an amplification with the URR and E6 primers and no amplification with the E2 primers in 15 out of 16 CIN III lesions and 19 out of 19 tumours (Table 1). In contrast, in CIN I\II lesions all three primer sets amplified the HPV-16 DNA fragments. Fig. 1 (a) represents an RNA in situ hybridizat...