Background: We examined the real-world effectiveness of ART as an HIV prevention tool among HIV serodiscordant couples in a programmatic setting in a low-income country. Methods: We enrolled individuals from HIV serodiscordant couples aged ≥18 years of age in Jinja, Uganda from June 2009 - June 2011. In one group of couples the HIV positive partner was receiving ART as they met clinical eligibility criteria (a CD4 cell count ≤250 cells/μL or WHO Stage III/IV disease). In the second group the infected partner was notyet ART-eligible. We measured HIV incidence by testing the uninfected partner every three months. We conducted genetic linkage studies to determine the source of new infections in seroconverting participants. Results: A total of 586 couples were enrolled of which 249 (42%) of the HIV positive participants were receiving ART at enrollment, and an additional 99 (17%) initiated ART during the study. The median duration of follow-up was 1.5 years. We found 9 new infections among partners of participants who had been receiving ART for at least three months and 8 new infections in partners of participants who had not received ART or received it for less than three months, for incidence rates of 2.09 per 100 person-years (PYRs) and 2.30 per 100 PYRs, respectively. The incidence rate ratio for ART-use was 0.91 (95% confidence interval 0.31-2.70; p=0.999). The hazard ratio for HIV seroconversion associated with ART-use by the positive partner was 1.07 (95% CI 0.41-2.80). A total of 5/7 (71%) of the transmissions on ART and 6/7 (86%) of those not on ART were genetically linked. Conclusion: Overall HIV incidence was low in comparison to previous studies of serodiscordant couples. However, ART-use was not associated with a reduced risk of HIV transmission in this study.