BACKGROUND: Protein p16(INK4a) immunocytochemistry (ICCp16) has the potential to reveal lesions at risk of progression to anal cancer. This study examined measures of diagnostic validity of ICCp16 in HIV-positive patients treated at the Tropical Medicine Foundation of Amazonas in the coloproctology outpatient clinic. METHODS: One hundred ninety HIV-positive patients were consecutively enrolled in 2007 and 2008. All patients underwent anal cytologic sampling to perform ICCp16 in conventional and GluCyte (Synermed International, Westfield, Indiana and S¸ao Paulo, Brazil) smears and also for genotyping of human papillomavirus (HPV). Patients were then subjected to anal biopsies monitored by high-resolution anoscopy. Hematoxylin-eosin and immunoperoxidase p16 (clone 6H12) stains were performed in slides with biopsied and cytological specimens, respectively. HPV genotyping on anal scrapings was performed by a polymerase-chain reaction (PCR)-based method. The immunochemical findings were compared with histopathological and PCR results in contingency tables and analyzed by nonparametric tests. Measures of diagnostic validity of ICCp16 were calculated. Statistical significance was set at P ≤ .5. RESULTS: There was no statistically significant association between the immunochemical results (conventional or GluCyte smears) and histopathological or HPV genotyping findings (P > .05). In the best scenario, ICCp16 presented 31% sensitivity and 81% specificity for the diagnosis of anal squamous intraepithelial lesion (ASIL) and 30% and 66%, respectively, for the diagnosis of infection with high-risk HPV. CONCLUSIONS: There was no association between ICCp16 results and histopathological findings nor between ICCp16 and HPV genotyping. ICCp16 showed poor sensitivity and moderate specificity for the diagnosis of ASIL or high-risk HPV.