BACKGROUND: We examined several measures of self-reported HIV risk behaviour in mutually disclosed sero-discordant couples over time to see if a couples counselling intervention was associated with changes in these behaviors. METHODS: We analysed data from a prospective cohort study of HIV sero-discordant couples in Jinja, Uganda collected between June 2009 and December 2011. Participants received couples counselling, at 3-monthly intervals. We examined trends in reported condom-use, number of concurrent sexual partners, knowledge of HIV serostatus of concurrent partners and condom use of concurrent partners using Generalized Estimating Equation models, comparing responses at study enrollment with responses at six, 12 18 and 24 months of follow-up. RESULTS: A total of 586 couples were enrolled and the female member was HIV positive in 255 (44%) of them. The median age for female participants was 35 years and 42 years for men. Reported condom use at last sex with spouse increased over time (p<0.001) with the largest increases found among couples where the positive participant never received ART during the study(an increase from 68.8% at enrollment to 97.1% at 24 months). Male participants reported reductions in the number of concurrent sexual partners (p<0.001), increase in the knowledge of the HIV serostatus of these partners (p = 0.001) and a trend towards improved condom-use among non-primary partners (p = 0.070). Reported reduced risky behaviors did not wane over the study period. CONCLUSION: Couples counselling resulted in increased condom use among all participants and among men the intervention resulted in reductions in risk behaviour with concurrent sexual partners. Routine counselling for serodiscordant couples should be integrated in routine ART care programs.