OBJECTIVE:To determine whether reporting that the HIV-positive partner's viral load is undetectable rather than detectable is associated with unprotected anal intercourse (UAI) in HIV serodiscordant gay couples. METHOD:A cross-sectional study nested within two cohort studies, the Health in Men (HIM) cohort of HIV-negative men, from July 2001 to December 2003 and the Positive Health (PH) cohort of HIV-positive men, from February 2002 to August 2003. The study participants were 119 men in an HIV serodiscordant regular relationship of at least 6 months duration (45 HIV-negative men from HIM, 74 HIV-positive men from PH). The main outcome measure was the occurrence of UAI within the relationship in the previous 6 months. RESULTS:Eighty-two men reported no UAI and 37 reported some UAI. Of couples in which the HIV-positive partner's viral load was reported to be undetectable, 39.4% reported UAI compared with 20.8% of those where viral load was reported to be detectable (P = 0.04). In multivariate analysis, significant predictors of UAI were younger age [odds ratio (OR), 0.94; 95% confidence interval (CI), 0.87-1.00; P = 0.05], greater HIV optimism (OR, 4.98; 95% CI, 1.25-19.8; P = 0.02) and reported undetectable viral load (OR, 2.88; 95% CI, 1.13-7.37; P = 0.03). CONCLUSIONS:Most serodiscordant gay couples do not engage in any UAI. UAI within such relationships is significantly more likely to occur where the HIV-positive partner is reported to have undetectable viral load. UAI in HIV serodiscordant relationships is problematic even if viral load is undetectable because of unknown risk parameters, viral load variability and the possibility of drug-resistant strains of HIV.