From September 2008 to December 2011, we enrolled and followed-up 247 HIV-negative, 88 untreated and 32 treated HIV-positive female sex workers (FSWs), as well as 238 untreated and 115 treated HIV-positive patients from the general population (GP) of Cotonou, Benin. We wanted to assess the effect of antiretroviral therapy (ART) on sexual risk-taking in FSWs and patients from the GP. We used multivariate log binomial regression models for repeated measures to compare risky behaviours reported during pre-ART and post-ART visits and we performed linear time-trend analyses to assess changes in condom use in all five groups. At 58.8% of pre-ART and 45.3% of post-ART visits (adjusted p-value=0.293), treated FSWs have reported ≥16 clients during the last week of work. Inconsistent condom use with clients over the same period decreased by more than 50% (from 20.7 to 10.0%, adjusted p-value=0.082). In treated patients from the GP, inconsistent condom use with regular partners during the last four months was reported at 52.8% of pre-ART and 53.5% of post-ART visits (p=0.778). Reported casual sex was stable (36.8% versus 38.7%, adjusted p-value=0.924). In linear time-trend analyses, there was a significant downward trend in inconsistent condom use at the early stage of the study and stability thereafter in all HIV-negative and HIV-positive FSWs. There was no negative alteration in sexual behaviour following ART initiation either inpatients from the GP or in FSWs. The results underscore the key role of concomitant sexual risk-reduction strategies.