OBJECTIVE The potential adverse effects of antiretroviral drugs during pregnancy are discrepant and few studies, mostly from Europe, have provided information about pregnancy outcomes of those already on treatment at conception. The aim of this study was to investigate the impact of antiretrovirals (ARVs) on pregnancy outcome according to the timing of treatment initiation in a cohort of pregnant women from Brazil infected with HIV. METHODS A prospective cohort of 696 pregnant women followed up in one single centre between 1996 and 2006 was studied. Patients who had ARV treatment before pregnancy were compared with those treated after the first trimester. The outcomes evaluated were preterm delivery (PTD) (<37 weeks), severe PTD (<34 weeks), low birth weight (LBW) (<2500 g) and very LBW (<1500 g). RESULTS Patients who were using ARVs pre-conception had higher rates of LBW (33.3% vs 16.5%; p<0.001) and a similar trend for PTD (26.3% vs 17.7%; p = 0.09). Stratification by type of therapy (dual vs highly active antiretroviral therapy (HAART)) according to timing of initiation of ARVs showed that patients who use HAART pre-conception have a higher rate of PTD (20.2% vs 10.2%; p = 0.03) and LBW (24.2% vs 10.2%; p = 0.002). After adjusting for several factors, HAART used pre-conception was associated with an increased risk for PTD (AOR 5.0; 95% CI 1.5 to 17.0; p = 0.009) and LBW (OR 3.6; 95% CI 1.7 to 7.7; p = 0.001). CONCLUSIONS We identified an increased risk for LBW and PTD in patients who had HAART prior to pregnancy.