Due to the high prevalence of cancer-associated types of human papillomavirus (HPV) and the poorly reproducible histologic classification of low-grade lesions, identifying infected women at highest risk for cancer prior to neoplastic progression remains a challenge. We therefore explored the utility of p16INK4a immunostaining as a potential diagnostic and prognostic biomarker for cervical neoplasia using paraffin-embedded tissue blocks (punch biopsies and loop electrosurgical excision procedures) obtained from women referred to colposcopy during the enrollment phase of the Guanacaste Project (1993 to 1994). All blocks from 292 women selected by HPV status (HPV negative, nononcogenic HPV positive, or oncogenic HPV positive) and representing the diagnostic spectrum of the population [normal to precancer: cervical intraepithelial neoplasia (CIN) 3] were immunostained for p16INK4a using the p16INK4a research kit based on the monoclonal antibody clone E6H4 (MTM Laboratories, Heidelberg, Germany). For CIN3, the sensitivity of diffuse p16INK4a immunostaining was 100% and the specificity was 95%. For CIN2, the sensitivity and specificity for diffuse staining were 81.1% and 95.4%, respectively. Generalized to the 10,000-woman cohort, this translated to positive predictive value and negative predictive value of 13.9% and 100% for CIN3, respectively, and 20.4% and 99.7% for CIN2 or CIN3, respectively. Of women with an initial diagnosis of less than CIN2 for whom follow-up data for up to 5 to 7 years were available, 44% with diffuse staining developed persistent infection (CIN2 or CIN3). Whereas our data support the diagnostic potential for p16INK4a, further prospective studies with detailed follow-up determining the prognostic capacity of this marker are needed.