Background Perinatal growth abnormalities program susceptibility to childhood obesity, which is further exaggerated by maternal overweight and obesity (MO) during pregnancy. Exercise is highly accessible, but reports about the benefts of maternal exercise on fetal growth and childhood obesity outcomes are inconsistent, reducing the incentives for pregnant women to participate in exercise to improve children’s perinatal growth. Objective This systematic review and meta-analysis aims to establish evidence-based efcacy of exercise in mothers with normal weight (MNW) and MO during pregnancy in reducing the risks of perinatal growth abnormalities and childhood obesity. In addition, the impacts of exercise volume are also assessed. Methods The PubMed, ScienceDirect, Web of Science, and Cochrane Library databases were searched from inception to February 15, 2020. We included randomized controlled trials with exercise-only intervention or exercise with other confounders in pregnant MNW (body mass index, BMI 18.5–24.9 kg/m2 ) and MO (BMI≥25 kg/m2 ), which were further subgrouped in the meta-analysis. Primary outcomes included birth weight, preterm birth, small for gestational age (SGA), large for gestational age (LGA), infant and childhood weight, and childhood obesity. A linear meta-regression analysis was also used to explore the efects of exercise volume on outcomes. Results 99 studies were included in the meta-analysis (n=596,876), and individual study quality ranged from fair to good according to the Newcastle–Ottawa scale assessment. Exercise only interventions in MNW reduced preterm birth by 15% (26 studies, n=76,132; odds ratio [OR] 0.85; 95% CI 0.72, 1.01; I 2=83.3%), SGA by 17% (33 studies, n=92,351; OR 0.83; 95% CI 0.71, 0.98; I 2=74.5%) and LGA by 17% (29 studies, n=84,310; OR 0.83; 95% CI 0.74, 0.95; I 2=60.4%). Exercise only interventions in MO reduced preterm birth by 33% (2 studies, n=3,050; OR 0.67; 95% CI 0.70, 0.96; I 2=0%), SGA by 27% (8 studies, n=3,909; OR 0.73; 95% CI 0.50, 1.05; I 2=40.4%) and LGA by 55% (9 studies, n=81,581; OR 0.45; 95% CI 0.18, 1.11; I 2=98.3%). Exercise only interventions in MNW reduced childhood obesity by 53% (3 studies, n=6,920; OR 0.47; 95% CI 0.36, 0.63; I 2=77.0%). However, no signifcant efect was observed in outcomes from exercise confounders in either MNW or MO. In the meta-regression, the volume of exercise-only intervention in MNW was negatively associated with birth weight, greatly driven by volumes more than 810 metabolic equivalents (MET)-min per week. Other outcomes were not associated with exercise volume. Conclusions This systematic review and meta-analysis suggests that exercise during pregnancy in both MNW and MO safely and efectively reduce the risks of preterm birth, SGA, and LGA. Furthermore, MNW exercise also reduces the risk of childhood obesity. Overall, regardless of prepregnancy BMI, maternal exercise during pregnancy provides an excellent opportunity to mitigate the high prevalence of adverse birth outcomes and childhood obesity