OBJECTIVE: To identify factors associated with mother-to-child HIV-1 transmission (MTCT) from mothers receiving antenatal antiretroviral therapy. DESIGN: The French Perinatal Cohort (EPF), a multicenter prospective cohort of HIV-infected pregnant women and their children. METHODS: Univariate analysis and logistic regression, with child HIV status as dependent variable, were conducted among 5271 mothers who received antiretroviral therapy during pregnancy, delivered between 1997 and 2004 and did not breastfeed. RESULTS: The MTCT rate was 1.3% [67/5271; 95% confidence interval (CI), 1.0-1.6]. It was as low as 0.4% (5/1338; 95% CI, 0.1-0.9) in term births with maternal HIV-1 RNA level at delivery below 50 copies/ml. MTCT increased with viral load, short duration of antiretroviral therapy, female gender and severe premature delivery: 6.6% before 33 weeks versus 1.2% at 37 weeks or more (P < 0.001). The type of antiretroviral therapy was not associated with transmission. Intrapartum therapy was associated with four-fold lower MTCT (P = 0.04) in case of virological failure (> 10 000 copies/ml). Elective cesarean section tended to be inversely associated with MTCT in the overall population, but not in mothers who delivered at term with viral load < 400 copies/ml [odds ratio (OR), 0.83; 95% CI, 0.29-2.39; P = 0.37]. Among them, only duration of antenatal therapy was associated with transmission (OR by week, 0.94; 95% CI, 0.90-0.99; P = 0.03). CONCLUSIONS: Low maternal plasma viral load is the key factor for preventing MTCT. Benefits in terms of MTCT reduction may be expected from early antiretroviral prophylaxis. The potential toxicity of prolonged antiretroviral use in pregnancy should be evaluated.