BACKGROUND: The World Health Organization recommends antiretroviral therapy (ART) for pregnant women living with HIV (WLHIV), the vast majority of whom reside in sub-Saharan Africa. In recent years, many systematic reviews, meta-analyses, and randomised controlled trials (RCTs) have been performed to assess the risks of adverse perinatal outcomes associated with ART among WLHIV. The aim of this umbrella review is to assess the available evidence regarding the risks of adverse perinatal outcomes for WLHIV receiving ART. METHODS: We conducted a systematic literature review by searching Medline, Global Health, and EMBASE and two clinical trial databases (ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform) for meta-analyses and RCTs published between 1 January 1980 and 12 February 2026. We included meta-analyses and RCTs reporting on the association of pregnant WLHIV receiving ART with perinatal outcomes, compared to WLHIV receiving different ART regimens, WLHIV naïve to ART, and women without HIV. We also included studies that assessed the timing of ART initiation (preconception or antenatal). Twelve predefined perinatal outcomes were assessed: preterm birth (PTB), very PTB (VPTB), spontaneous PTB (sPTB), low birthweight (LBW), very LBW (VLBW), term and preterm LBW, small for gestational age (SGA), very SGA (VSGA), stillbirth, neonatal death, and vertical HIV transmission. Heterogeneity measures (I 2 ) and Peter's test results for publication bias were extracted and assessed. The quality of meta-analyses was assessed using the AMSTAR 2 tool, risk of bias of RCTs was assessed using the Cochrane Risk of Bias tool, and overall certainty of evidence for each exposure comparison and perinatal outcome was assessed according to GRADE. The protocol is registered with PROSPERO, number CRD42021248987. FINDINGS: Of 14,279 studies identified, we included nine meta-analyses of cohort studies, one network meta-analysis of seven RCTs, and three additional RCTs. The meta-analyses were composed of a total of 154 cohort studies and were of low or critically low quality, and RCTs had a high risk of bias. Meta-analyses of cohort studies found that WLHIV receiving ART are at increased risks of PTB (risk ratio (RR) 1.55, 95% confidence interval (CI) 1.38-1.74, I 2 87.4%), sPTB (RR 2.10, 95% CI 1.48-2.96, I 2 12.5%), LBW (RR 1.79, 95% CI 1.51-2.13, I 2 90.6%), term LBW (RR 1.88, 95% CI 1.23-2.85, I 2 0.0%), SGA (RR 1.80, 95% CI 1.34-2.40, I 2 97.6%), and VSGA (RR 1.22, 95% CI 1.10-1.34, I 2 0.0%), compared to HIV-negative women. RCTs comparing ART regimens among WLHIV found few differences in perinatal outcomes assessed. Meta-analyses of cohort studies comparing different ART regimens found that protease inhibitors are associated with an increased risk of SGA (RR 1.28, 95% CI 1.09-1.51, I 2 62.2%) and VSGA (RR 1.41, 95% CI 1.08-1.83, I 2 0.0%), compared to non-nucleoside reverse transcriptase inhibitors. Tenofovir disoproxil fumarate is associated with a lower risk of adverse perinatal outcomes and zidovudine is associated with an increased risk of perinatal outcomes. Compared to HIV-negative women, WLHIV receiving ART remain at increased risk of adverse perinatal outcomes, irrespective of the ART regimen and timing of ART initiation. Publication bias, as determined using the Peter's test, was found for two analyses. Most findings were of low or very low certainty as assessed using GRADE. INTERPRETATION: WLHIV receiving ART are associated with increased risks of adverse perinatal outcomes compared to HIV-negative women, irrespective of the ART regimen and timing of ART initiation. To strengthen the evidence base for ART guidelines for pregnant WLHIV, more and larger RCTs and high quality observational studies are needed to optimise ART regimens in pregnancy. Further efforts should be made to improve perinatal outcomes among WLHIV. FUNDING: This study received no funding.