Perinatal HIV transmission has substantially decreased with combination antiretroviral regimens, but complications in children who are HIV-exposed but uninfected, such as microcephaly, warrant ongoing surveillance. We aimed to evaluate whether individual in utero antiretroviral exposures were associated with increased risk of microcephaly based on long-term follow-up of infants and children who are HIV-exposed but uninfected.We evaluated children aged younger than 18 years who were HIV-exposed but uninfected with at least one head circumference measurement while enrolled in the Surveillance Monitoring for ART Toxicities (SMARTT) study at 22 clinical sites in the USA, including Puerto Rico. This prospective cohort study was done by the Pediatric HIV/AIDS Cohort Study network. Microcephaly was defined as having a head circumference Z score-2 according to the 2000 US Centers for Disease Control and Prevention growth charts for children 6-36 months old and according to Nellhaus standards (head circumference2nd percentile) after 36 months (SMARTT criteria); an alternate definition for microcephaly was based on applying Nellhaus standards across all ages (Nellhaus criteria). Modified Poisson regression models were fit to obtain relative risks (RRs) for associations between in utero antiretroviral exposure and microcephaly status, adjusted for potential confounders. Neurodevelopmental functioning was compared in children who are HIV-exposed but uninfected with or without microcephaly.Between March 21, 2007, and Aug 1, 2017, 3055 participants enrolled in SMARTT had at least one head circumference measurement. The cumulative incidence of microcephaly over a median of 5·1 years of follow-up (IQR 3·0-7·2) was 159 (5·2%, 95% CI 4·4-6·1) by Nellhaus criteria and 70 (2·3%, 1·8-2·9) by SMARTT criteria. In adjusted models, in utero exposure to efavirenz (4·7% exposed) was associated with increased risk of microcephaly by both Nellhaus standards (adjusted RR 2·02, 95% CI 1·16-3·51) and SMARTT criteria (2·56, 1·22-5·37). These associations were more pronounced in children exposed to combination regimens of efavirenz that included zidovudine plus lamivudine than those including tenofovir plus emtricitabine. Protective associations were observed for darunavir exposure (adjusted RR 0·50, 95% CI 0·24-1·00). Children who are HIV-exposed but uninfected with microcephaly had lower mean scores on neurodevelopmental assessments at age 1 and 5 years and a higher prevalence of neurodevelopmental impairment than those without microcephaly.These findings support consideration of alternatives to efavirenz as part of first-line antiretroviral therapy for pregnant women.Eunice Kennedy Shriver National Institute of Child Health and Human Development.