OBJECTIVE: To assess the long-term effects of population-level HSV-2 infection on HIV incidence. METHODS: Data from a population-based cohort in south-western Uganda were used to estimate HIV incidence from 1990 to 2007. Stored blood samples were tested for HSV-2, and the impact of HSV-2 prevalence and incidence on HIV incidence was estimated by calculating population attributable fractions (PAFs). The association between population-level annual HIV incidence and annual HSV-2 incidence/prevalence was analysed using linear regression. RESULTS: HIV incidence declined over time among men, from 8.72/1000 person-years (pyr) in 1990 to 4.85/1000 pyr in 2007 (P-trend <0.001). In contrast, there was no decline in HIV incidence among women (4.86/1000 pyr in 1990 to 6.74/1000 pyr in 2007, P-trend = 0.18). PAFs of incident HIV attributable to HSV-2 were high (60% in males; 70% in females). There was no evidence of an association between long-term trends in HIV incidence and HSV-2 prevalence or incidence. CONCLUSION: Assuming a causal relationship, a substantial proportion of new HIV infections in this population are attributable to HSV-2. The study did not find an effect of HSV-2 prevalence/incidence on trends in HIV incidence. HIV incidence did not vary much during the study period. This may partly explain the lack of association.