BACKGROUND: There are 15·4 million children who are HIV-exposed and uninfected worldwide. Early child development crucially influences later academic and socioeconomic factors. However, the neurodevelopmental outcomes of HIV-exposed uninfected (HEU) children in the era of maternal antiretroviral therapy (ART) remain unclear. We aimed to examine the effects of in-utero exposure to HIV and ART on child neurodevelopment. METHODS: For this systematic review and meta-analysis, we searched MEDLINE, Embase, PubMed, Africa-Wide Information, PsycInfo, and Global Health databases from inception to May 27, 2020, for studies from the past two decades reporting neurodevelopment of HEU children aged 0-5 years compared with HIV-unexposed (HU) children (aim 1), and effects of different maternal ART regimens on neurodevelopment of HEU children (aim 2). We did narrative syntheses for both aims, and a random-effects meta-analysis of high-quality studies comparing HEU children and HU children, to obtain weighted pooled estimates of effect sizes. This study was registered with PROSPERO, CRD42018075910. FINDINGS: We screened 35 527 records and included 45 articles from 31 studies. Overall, 12 (57%) of 21 studies comparing HEU children and HU children found worse neurodevelopment in HEU children in at least one domain. Study design and methodological quality were variable, with heterogeneity across populations. Meta-analysis included eight high-quality studies comparing 1856 HEU children with 3067 HU children at ages 12-24 months; among HEU children with available data, 1709 (99%) of 1732 were exposed to ART. HEU children had poorer expressive language (effect size -0·17 [95% CI -0·27 to -0·07], p=0·0013) and gross motor function (-0·13 [-0·20 to -0·07], p<0·0001) than HU children, but similar cognitive development (-0·06 [-0·19 to 0·06], p=0·34), receptive language development (-0·10 [-0·23 to 0·03], p=0·14), and fine motor skills (-0·05 [-0·15 to 0·06], p=0·36). Results suggested little or no evidence of an effect of specific maternal ART regimens on neurodevelopment; study heterogeneity prevented meta-analysis. INTERPRETATION: HEU children are at risk of subtle impairments in expressive language and gross motor development by age 2 years. We found no consistent effect of maternal ART regimens analysed, although evidence was scarce. We highlight the need for large high-quality longitudinal studies to assess the neurodevelopmental trajectories of HEU children and to investigate underlying mechanisms to inform intervention strategies. FUNDING: Wellcome Trust and Medical Research Council.