To integrate the empirical findings on the prevalence of unprotected anal intercourse (UAI) among HIV-diagnosed men who have sex with men (MSM) in the United States. Comprehensively searching MEDLINE, EMBASE, PsycINFO (2000-2007), hand searching bibliographic lists, and contacting researchers. Thirty US studies (n = 18,121) met selection criteria. Analyses were conducted using random-effects models and meta-regression. The prevalence of UAI was considerably higher with HIV-seropositive partners (30%; 95% confidence interval 25-36) than with serostatus unknown (16%; 95% confidence interval 13-21) or HIV-seronegative partners (13%; 95% confidence interval 10-16). The prevalence of UAI with either a serostatus unknown or HIV-seronegative partner was 26%. The UAI prevalence did not differ by the length of the behavioral recall window but did vary by the type of anal intercourse (insertive vs. receptive). Studies with the following features had a lower UAI prevalence: recruiting participants before 2000, MSM of color being the majority of study sample, recruiting participants from medical settings, using random or systematic sampling methods, and having interviewers administer the questionnaire. Being on antiretroviral therapy, having an undetectable viral load, and reporting more than 90% medication adherence were not associated with UAI. Most HIV-diagnosed MSM protect partners during sexual activity, but a sizeable percentage continues to engage in sexual behaviors that place others at risk for HIV infection and place themselves at risk for other sexually transmitted infections. Prevention with positives programs continues to be urgently needed for MSM in the United States.
