Summary Background Carbon dioxide (CO2) insufflation has been proposed as an alternative to air insufflation to distend the lumen in gastrointestinal (GI) endoscopy. Aim To perform a systematic review with meta‐analysis of randomised controlled trials (RCTs) in which CO2 insufflation was compared with room air insufflation in GI endoscopy. Methods Electronic and manual searches were combined to search RCTs. After methodological quality assessment and data extraction, the efficacy and safety of CO2 insufflation were systematically assessed. Results Twenty‐one RCTs 13 on colonoscopy, four on endoscopic retrograde cholangiopancreatography (ERCP), two on double‐balloon enteroscopy (DBE), one on oesophagogastroduodenoscopy, and one on flexible sigmoidoscopy were identified. For colonoscopy, CO2 insufflation resulted lower postprocedural pain intensity, and increased the proportion of patient without pain at 1 h (RR: 1.84, 95% CI: 1.37–2.47) and 6 h (RR: 1.28; 95% CI: 1.14–1.44) postprocedure. For ERCP, the pain‐releasing effect of CO2 insufflation was not obvious (SMD: −1.48, 95% CI: −3.56, 0.59). CO2 insufflation revealed no consistent advantages in the RCTs of DBE, but was shown as safe as air insufflation in oesophagus/stomach endoscopic submucosal dissection in one study. pCO2 level showed no significant variation during these procedures. Conclusions Compared with air insufflation, CO2 insufflation during colonoscopy causes lower postprocedural pain and bowel distension without significant pCO2 variation. More RCTs are needed to assess its advantages in other GI endoscopic procedures.