To evaluate the effect of extended antenatal triple antiretroviral therapy (ART) on infant outcomes. Retrospective cohort study using pooled data from health clinics in Malawi and Mozambique from July 2005 to December 2009. Computerized records of 3273 HIV-infected pregnant women accessing Drug Resource Enhancement Against AIDS and Malnutrition centers were reviewed. ART regimens consisted of nevirapine-based HAART as of 14-25 weeks gestation until 6 months postpartum. Infant infection was determined at 1, 6 and 12 months of age by branched DNA. A total of 3071 pregnancies resulted in 3148 live births. Lost to follow-up, infant deaths and HIV-1 infection rates at 1 and 12 months were 1.3 and 11.5, 0.8 and 6.7 and 0.8 and 2.0, respectively. Infant HIV-1-free survival at 12 months was 92.5%. Mother-to-child transmission and/or infant deaths correlated with length of maternal antenatal ART by multivariate analysis at 1, 6 and 12 months: 14% in women with more than 30 days of triple antenatal ART and 6.9% in mothers receiving at least 90 days of antenatal ART, P = 0.001. Fifty percent of 54 episodes of transmission occurred in women with higher CD4 cell counts (>350 cells/μl). Infant mortality was 67/1000, lower than background rates (78-100/1000). Growth failure (weight-for-age Z score <-2) was present in 8% of infants around birth, 6% at 6 months, 23% at 12 months (lower than country-specific rates). Extended antenatal ART is protective against adverse infant outcomes up to 12 months of age even in children born to mothers with higher CD4 cell counts.