Background: Before the completion of HPTN 052, PARTNER, and Opposites Attract studies, data were lacking to directly estimate HIV transmission risk under effective combined antiretroviral treatment (cART). Rather, estimates were obtained by extrapolating a dose-response relationship between viral load and risk of HIV transmission, observed among untreated individuals, to treated individuals. Presently, data have accumulated from 9 clinical studies for a direct validation of this extrapolation. Methods: Using estimates of per sex-act risk of HIV transmission on effective cART obtained by extrapolation, sexual behavior data, and a simple mathematical model, we estimated the number of seroconversions that should have been observed in HIV-serodiscordant couples where the HIV-positive partner was on cART across the 9 studies. We compared this with the number of seroconversions actually observed. Next, we directly estimated the risk of HIV transmission on effective cART, using Bayesian statistics to combine all available data. Results: We found that at least 4.7 (uncertainty bounds: 1.7-12.6) and 35.1 (uncertainty bounds: 13.2-92.0) seroconversions should have been observed among, respectively, heterosexual and men who have sex with men (MSM) serodiscordant couples. This is not validated by observations across the studies, which reported at most 1 seroconversion among heterosexual couples and 0 for MSM. Combining all available data, we found that the maximum per sex-act risk of HIV transmission under effective cART is 3.9:100,000 for heterosexuals and 4.4:100,000 for MSM. Conclusions: Data have accumulated to render obsolete estimates of the risk of HIV transmission on cART obtained by extrapolation. Direct estimates are substantially lower and should be used in practice.