Objective Explore the differences in effectiveness and safety between endoscopic sub- mucosal dissection （ESD） and endoscopic mucosal resection （EMR） for early esophageal cancer （EEC）. Methods Papers published from January 1990 to December 2012 in Medline, Embase, Cochrane library, Wanfang, VIP and CNKI were searched in English and Chinese respectively to identify the studies comparing ESD with EMR for EEC. Meta-analysis for each outcome of the included studies that were identified by the inclusion criteria and the exclusion criteria was conducted using software Revman 5.1. Outcome measures consisted of three effectiveness-relevant measures （ rates of en bloc resection, curative resection, local recur- rence） and four safety-relevant measures （bleeding, perforation, stenosis and operation time）. Results A total of 8 non-randomized retrospective cohort studies （five full-text and three abstracts） were included. Me- ta-analysis showed significantly higher rates of en bloc resection [ 98.36% （560/366） vs. 41.79% （252/603）, P 〈0. 01 ], curative resection [90. 81% （168/185） vs. 50. 65% （ 194/383）, P 〈0. 01 ] and perforation [g. 51% （21/466） vs. 1.25% （8/640）, P =0.05] in ESD group compared with EMR group, and the rate of local recurrence was significantly lower in ESD group [ 0. 55% （2/366） vs. 13.76% （85/603）, P 〈0.01 ]. There were no significant differences in the rates of bleeding [0.21% （1/466） vs. 0.63%（4/640）, P =0.41] or stenosis [10.48% （39/372） vs. 10.15% （41/404）, P =0.89].Conclusion Compared with EMR, ESD showed obvious advantages of effectiveness and resemble risks for endoscopic resection of EEC, therefore ESD should be the first choice for endoscopic treatment of EEC.