Although there is consensus that HPV integration is common in invasive cervical carcinomas and uncommon or absent in low-grade uterine cervical intraepithelial neoplasia (CIN I), estimates for HPV integration in CIN II/III range from 5 to 100% using different PCR-based and in situ hybridization (ISH) approaches. It has been suggested that HPV integration can be identified using ISH by scoring of punctate signals. The increased sensitivity of fluorescence ISH (FISH) methods, allowing the detection of single copies of HPV, complicates the distinction between integrated and episomal HPV. Recently it has been suggested that, in such assays, the signals originating from integrated virus can be hidden in a background of episomal HPV. We therefore compared 2 different FISH protocols for the detection of integrated HPV in a series of CIN II/III lesions: 1) a mild protocol in which episomal HPV and RNA is retained and 2) a harsh protocol that extensively extracts proteins and RNA, and which promotes the partial loss of episomal HPV but not integrated HPV. A series of 28 HPV 16/18 positive CIN II/III lesions (17 solitary lesions and 11 lesions adjacent to microinvasive carcinoma) were studied. A punctate signal pattern was identified in 7 of these lesions with both protocols. Punctate signal was also present in control samples from lesions that are known to be associated with HPV integration (invasive squamous cell carcinoma (n = 3), adenocarcinoma in situ (n = 3), and invasive adenocarcinoma (n = 1). HPV RNA contributed significantly to the intensity of punctate FISH signal, especially when applying the mild protocol, as shown by omitting DNA denaturation, including RNase pretreatment steps and measuring the fluorescence signal intensity. Also, HPV RNA was frequently detected in addition to episomal/integrated HPV DNA in the majority of the other 21 CIN II/III lesions; this resulted in intense granular/diffuse FISH signals throughout the epithelium. However, in 7 of these lesions, the harsh protocol gave a more consistent punctate pattern in cells throughout the full thickness of the epithelium. This supports the hypothesis that the harsh protocol unmasks integrated HPV more efficiently by extracting RNA and episomal HPV. Overall, with this harsh protocol, a clonally expanded population of cells containing punctate HPV signals was found in 5 of 17 (29%) solitary CIN II/III lesions and in 9 of 11 (88%) CIN II/III lesions associated with microinvasive carcinoma. Combining these data with the results from our previous study, with the harsh protocol in 7 of 40 (18%) solitary CIN II/III lesions and 19/21 (90%) CIN II/III lesions associated with microinvasive carcinoma (p < 0.001), this pattern was found. This indicates that, when robustly defined, a punctate HPV pattern in CIN II/III lesions is associated with the presence of an invasive carcinoma. © 2005 Wiley-Liss, Inc.