OBJECTIVE: To examine the incidence and risk factors for anal cancer in a multicenter cohort of human immunodeficiency virus (HIV) positive and HIV-negative men who have sex with men followed between 1984 and 2006 (Multicenter AIDS Cohort Study). METHODS: Prospective analysis using Poisson regression and Cox proportional hazard models and a nested case-control study using conditional logistic regression. RESULTS: There were 28 cases of anal cancer among the 6,972 men who were evaluated. The incidence rate was significantly higher in HIV-positive men than in HIV-negative men (incidence rate = 69 vs 14 per 100,000 person-years). Among HIV-positive men, anal cancer incidence was higher in the highly active antiretroviral therapy (HAART) era than the pre-HAART era (incidence rate = 137 vs 30 per 100,000 person-years). In multivariate analysis restricted to the HAART era, anal cancer risk increased significantly with HIV infection (relative hazard = 4.7, 95% confidence interval = 1.3 to 17) and increasing number of unprotected receptive anal sex partners at the first 3 study visits (P trend = 0.03). Among HIV-positive men, current HAART use did not decrease anal cancer risk. CONCLUSIONS: HIV-positive men had increased risk of anal cancer. Improved survival of HIV-positive individuals after HAART initiation may allow for sufficient time for human papillomavirus-associated anal dysplasias to develop into malignancies, thus explaining the increased incidence of anal cancer in the HAART era.