From the pathogenic point of view, penile cancers may be grouped in human papillomavirus-related and unrelated tumors, each one of them with distinctive morphologic features. The former are predominantly composed of small, undifferentiated basaloid cells, with more or less prominent koilocytic changes, and the latter of keratinizing differentiated squamous cells. The same cellular types are observed in precancerous lesions. On the basis of these observations, we constructed a novel nomenclature for penile precancerous lesions and classified them as penile intraepithelial neoplasia (PeIN) of differentiated, warty, basaloid, and warty-basaloid types. The aim of this study was to test the usefulness of immunohistochemical p16INK4a overexpression, considered as a surrogate for high-risk human papillomavirus infection, using this classification system. We pathologically evaluated 141 patients with PeIN, associated (123 cases) and unassociated (18 cases) with invasive cancer. Distribution of PeIN types was: differentiated, 72%; basaloid, 9%; warty-basaloid, 7%; warty, 4%; and mixed, 7%. There was a striking similarity in the morphology of in situ and invasive squamous cell carcinomas. Differentiated PeIN was commonly associated with usual, verrucous, papillary, and other low-grade keratinizing variants of squamous cell carcinoma whereas in basaloid and warty carcinomas the presence of in situ lesions with similar morphology was habitual. We evaluated p16INK4a overexpression using a 4-tiered (0, 1, 2, and 3) pattern-based system. To properly distinguish differentiated PeIN from in situ lesions with warty and/or basaloid features only pattern 3, which requires full-thickness staining in all epithelial cells, was considered positive. Using this approach, there was a significant association of the negative patterns and differentiated PeIN and of the positive pattern and warty, basaloid, and warty-basaloid PeIN (P<0.0001). Basaloid variant had the strongest association. The sensitivity rate of p16INK4a positivity for discriminating types of PeIN was of 82%, with a specificity of 100% and an accuracy of 95%. Lichen sclerosus was identified in 42 cases and their epithelial component was p16INK4a negative in all cases. Although more studies are necessary to confirm these observations, p16INK4a overexpression seems to be a useful tool for discriminating differentiated from warty, basaloid, and warty-basaloid PeIN.