BACKGROUND: Eliminating mother-to-child transmission of HIV (EMTCT), defined as ≤50 infant HIV infections per 100 000 live births, is a global priority. Since 2011 policies to prevent mother-to-child transmission of HIV (PMTCT) shifted from maternal antiretroviral (ARV) treatment or prophylaxis contingent on CD4 cell count to lifelong maternal ARV treatment (cART). We sought to measure progress with early (4-8 weeks postpartum) MTCT prevention and elimination, 2011-2013, at national and sub-national levels in South Africa, a high antenatal HIV prevalence setting ( ≈ 29%), where early MTCT was 3.5% in 2010. METHODS: Two surveys were conducted (August 2011-March 2012 and October 2012-May 2013), in 580 health facilities, randomly selected after two-stage probability proportional to size sampling of facilities (the primary sampling unit), to provide valid national and sub-national-(provincial)-level estimates. Data collectors interviewed caregivers of eligible infants, reviewed patient-held charts, and collected infant dried blood spots (iDBS). Confirmed positive HIV enzyme immunoassay (EIA) and positive total HIV nucleic acid polymerase chain reaction (PCR) indicated infant HIV exposure or infection, respectively. Weighted survey analysis was conducted for each survey and for the pooled data. FINDINGS: National data from 10 106 and 9120 participants were analyzed (2011-12 and 2012-13 surveys respectively). Infant HIV exposure was 32.2% (95% confidence interval (CI) 30.7-33.6%), in 2011-12 and 33.1% (95% CI 31.8-34.4%), provincial range of 22.1-43.6% in 2012-13. MTCT was 2.7% (95% CI 2.1%-3.2%) in 2011-12 and 2.6% (95% CI 2.0-3.2%), provincial range of 1.9-5.4% in 2012-13. HIV-infected ARV-exposed mothers had significantly lower unadjusted early MTCT (2.0% [2011-12: 1.6-2.5%; 2012-13:1.5-2.6%]) compared to HIV-infected ARV-naive mothers [10.2% in 2011-12 (6.5-13.8%); 9.2% in 2012-13 (5.6-12.7%)]. Pooled analyses demonstrated significantly lower early MTCT among exclusive breastfeeding (EBF) mothers receiving >10 weeks ARV prophylaxis or cART compared with EBF and no ARVs: (2.2% [95% CI 1.25-3.09%] vs 12.2% [95% CI 4.7-19.6%], respectively); among HIV-infected ARV-exposed mothers, 24.9% (95% CI 23.5-26.3%) initiated cART during or before the first trimester, and their early MTCT was 1.2% (95% CI 0.6-1.7%). Extrapolating these data, assuming 32% EIA positivity and 2.6% or 1.2% MTCT, 832 and 384 infants per 100 000 live births were HIV infected, respectively. CONCLUSIONS: Although we demonstrate sustained national-level PMTCT impact in a high HIV prevalence setting, results are far-removed from EMTCT targets. Reducing maternal HIV prevalence and treating all maternal HIV infection early are critical for further progress.