BACKGROUND: The global transition to use of dolutegravir (DTG) in WHO-preferred regimens for HIV treatment is limited by lack of knowledge on use in pregnancy. Here we assessed the relationship between drug concentrations (pharmacokinetics, PK), including in breastmilk, and impact on viral suppression when initiated in the third trimester (T3). METHODS AND FINDINGS: In DolPHIN-1, HIV-infected treatment-naïve pregnant women (28-36 weeks of gestation, age 26 (19-42), weight 67kg (45-119), all Black African) in Uganda and South Africa were randomised 1:1 to dolutegravir (DTG) or efavirenz (EFV)-containing ART until 2 weeks post-partum (2wPP), between 9th March 2017 and 16th January 2018, with follow-up until six months postpartum. The primary endpoint was pharmacokinetics of DTG in women and breastfed infants; secondary endpoints included maternal and infant safety and viral suppression. Intensive pharmacokinetic sampling of DTG was undertaken at day 14 and 2wPP following administration of a medium-fat breakfast, with additional paired sampling between maternal plasma and cord blood, breastmilk and infant plasma. No differences in median baseline maternal age, gestation (31 vs 30 weeks), weight, obstetric history, viral load (4.5 log10 copies/mL both arms) and CD4 count (343 vs 466 cells/mm3) were observed between DTG (n = 29) and EFV (n = 31) arms. Although DTG Ctrough was below the target 324ng/mL (clinical EC90) in 9/28 (32%) mothers in the third trimester, transfer across the placenta (121% of plasma concentrations) and into breastmilk (3% of plasma concentrations), coupled with slower elimination, led to significant infant plasma exposures (3-8% of maternal exposures). Both regimens were well-tolerated with no significant differences in frequency of adverse events (two on DTG-ART, one on EFV-ART, all considered unrelated to drug). No congenital abnormalities were observed. DTG resulted in significantly faster viral suppression (P = 0.02) at the 2wPP visit, with median time to <50 copies/mL of 32 vs 72 days. Limitations related to the requirement to initiate EFV-ART prior to randomisation, and to continue DTG for only two weeks postpartum. CONCLUSION: Despite low plasma DTG exposures in the third trimester, transfer across the placenta and through breastfeeding was observed in this study, with persistence in infants likely due to slower metabolic clearance. HIV RNA suppression <50 copies/mL was twice as fast with DTG compared to EFV, suggesting DTG has potential to reduce risk of vertical transmission in mothers who are initiated on treatment late in pregnancy. TRIAL REGISTRATION: clinicaltrials.gov NCT02245022.