BACKGROUND: In south India, general population HIV prevalence estimates range from 0.5 to 3%. To focus HIV prevention efforts, it is important to understand whether HIV transmission is driven by commercial sex. METHODS: A dynamic HIV/sexually transmitted infection transmission model was parameterized using data from Belgaum and Mysore in south India. Fits to sexually transmitted infection/HIV data from female sex workers (FSWs) and their clients for each district were obtained. Model HIV/herpes simplex virus-2 (HSV-2) prevalence projections for the general population were cross-validated against empirical estimates not used to fit model. The model estimated the proportion of incident HIV/HSV-2 infections due to HIV/HSV-2 transmission between FSWs/clients, their noncommercial partners and other low-risk partnerships. The relative impact of a generic intervention targeting different partnerships was explored. RESULTS: The model's general population HIV/HSV-2 prevalence projections agreed well with empirical estimates. Recent increases in condom use resulted in decreasing HIV epidemics in both settings. For men, most incident HIV/HSV-2 infections (>90%) directly result from commercial sex, whereas for women most are due to bridging infections from clients of FSWs (80-90%) with the remainder mainly due to commercial sex. Less than 1.5% of incident infections are due to low-risk partnerships. Intervention impact is maximized through targeting commercial sex but substantial impact could also be achieved through targeting noncommercial partners of clients. DISCUSSION: In southern India, HIV transmission could be driven by FSWs and their clients. While efforts to reduce HIV transmission due to commercial sex must continue, prevention programmes should also consider strategies to prevent transmission from clients to their noncommercial partners.