Background Robotic gastrectomy (RG) has been developed to improve surgical quality and to overcome the limitations of conventional open gastrectomy (OG) for gastric cancer. The aim of this meta-analysis is to comprehensively compare the safety and efficacy between robotic surgery and open surgery for treating gastric cancer. Methods Major databases were searched for retrospective case-matched studies comparing RG and OG for treating gastric cancer. A list of these studies, published in English from 1990 to 2016, was obtained independently by two reviewers from databases such as PubMed, MEDLINE, ScienceDirect, the China National Knowledge Infrastructure and Web of Science. Intraoperative data, oncological outcomes and postoperative complications were compared using Review Manager 5.3. Results Seven studies involving 5970 patients with 606 cases of RG and 5364 cases of OG were included in this meta-analysis. Compared to OG, RG has a significantly longer operation time [weighted mean differences (WMD) = 63.72, 95 % confidence interval (CI) 33.83-93.61, P < 0.0001], lower blood loss (WMD: -129.74, 95 % CI -178.31 to -81.16, P < 0.00001) and shorter hospital stay (WMD = -2.39, 95 % CI -2.92 to -1.87; P < 0.00001). No statistical difference was noted based on the rate of overall postoperative complication, wound infection, bleeding, ileus and obstruction, abdominal collections and abscesses, and the rate of anastomotic leak in the RG versus OG. Postoperative oncological outcomes showed that there were also no statistical differences among the number of retrieved lymph nodes, proximal resection margin, distal resection margin except for tumor size (WMD = -1.60; 95 % CI -2.96 to -0.25; P = 0.02). Conclusion The results of this meta-analysis suggest that RG will be more accessible than conventional OG for gastric cancer. However, more prospective, well-designed, multicenter, randomized controlled trials are necessary to further evaluate the safety and efficacy as well as the long-term outcome of this technology.