Endoscopic submucosal dissection (ESD) was originally developed for en bloc resection of large, flat gastrointestinal lesions. Compared with endoscopic mucosal resection (EMR), ESD is considered to be more time consuming and have more complications for treatment of early esophageal carcinoma, such as bleeding, stenosis and perforation. The objective of this study was to compare the efficacy and safety of ESD and EMR for such lesions. We searched databases, such as PubMed, EMBASE, Cochrane Library and Science Citation Index updated to 2013 for related trials. In the meta-analysis, the main outcome measurements were the en bloc resection rate, the histologically resection rate and the local recurrence rate. We also compared the operation time and the incidences of procedure-related complications. Five trials were identified, and a total of 710 patients and 795 lesions were included. The en bloc and histologically complete resection rates were higher in the ESD group compared with the EMR group (odds ratio (OR) 27.3; 95% CI, 11.5-64.8; OR 18.4; 95% CI, 8.82-38.59). The local recurrence rate was lower in the ESD group (OR 0.13, 95 % CI 0.04-0.43). The meta-analysis also showed ESD was more time consuming, but did not increase the complication rate (P=0.76). The results implied that compared with EMR, ESD showed better en bloc and histologically resection rates, and lower local recurrence, without increasing the incidence of procedure-related complications in the treatment of early esophageal carcinoma.