Background Epidemiological studies reported inconsistent results on the associations between maternal caffeine intake during pregnancy and risk of low birth weight (LBW) and childhood overweight and obesity in their offspring. Methods We conducted a meta-analysis of cohort studies to quantitatively assess these associations. Pertinent studies were identified by searching PubMed and Embase through June 2019. Study-specifics risk estimates were combined using fixed effects models, or random-effects models when significant heterogeneity was detected. Dose-response analysis was modeled by using restricted cubic splines. Results A total of 15 cohort studies, with 102,347 pregnancy women, was included in the meta-analysis. The pooled relative risk (RR) for LBW was 1.33 (95% CI: 1.12, 1.57) for mothers with the highest compared with the lowest level of caffeine intake during pregnancy, with significant heterogeneity across studies (I-2 = 49.3%,P = 0.032). The pooled RR was 1.07 (95% CI: 1.02, 1.11) for each 100 mg/day increase of caffeine intake. The pooled RR for childhood overweight and obesity was 1.39 (95% CI: 1.15, 1.69) for mothers with the highest compared with the lowest level of caffeine intake during pregnancy. No significant heterogeneity across studies was detected (I-2 = 38.9%,P = 0.179). The pooled RR was 1.31 (95% CI: 1.11, 1.55) for each 100 mg/day increase of caffeine intake. No evidence of publication bias was indicated. Conclusions Maternal caffeine intake during pregnancy is associated with higher risk of LBW and childhood overweight and obesity. Further studies may focus on investigating the potential mechanisms before the recommendation of complete avoidance of caffeine intake during pregnancy.