To compare HIV prevalence measured by antenatal clinics (ANC) sentinel surveillance and by the prevention of mother-to-child transmission (PMTCT) program in Rwanda. We compared HIV prevalence from anonymous testing performed under ANC surveillance, and that measured from voluntary counselling and testing performed under the PMTCT program, in a random sample of the same population of pregnant women attending for their first antenatal visit at 29 ANC surveillance sites with a PMTCT program in 2007 in Rwanda. All of the 13,318 pregnant women recruited in the ANC surveillance accepted to participate in the PMTCT program. HIV prevalence measured by sentinel surveillance was 4.35% whereas that measured for 1873 pregnant women (out of the total sentinel population) by the PMTCT program was 3.49% (p=0.07). For 3% of the PMTCT population, HIV test results were missing from the counselling logbook versus 0.3% in the ANC laboratory logbooks. For 10 pregnant women, HIV test results were divergent between the PMTCT and the ANC laboratory logbooks. After missing data and errors were corrected, HIV prevalence results from PMTCT was 3.27% (significantly different from ANC surveillance: p =0.03). High uptake of PMTCT program among pregnant women was observed in Rwanda in 2007. HIV prevalence measured by the ANC surveillance and PMTCT program were significantly different. Poor performance in HIV testing practices and PMTCT/laboratories data management could explain this difference. Improvement in HIV testing practices and in PMTCT/laboratory data management are needed in order to use PMTCT data for HIV surveillance and to ensure good performance of all the package of care provided by the PMTCT program.