Background and study aims: Controversy persists around the treatment of gastric low-grade dysplasia (LGD). The aim of this study was to investigate possible indications for the endoscopic resection of gastric LGD through analysis of the histologic discrepancies between specimens of gastric LGD obtained by forceps biopsy and by endoscopic mucosal resection (EMR), and of their clinicopathologic characteristics. Patients and methods: The study involved 293 gastric LGD that were histologically proven on the basis of forceps biopsy in Severance Hospital between January 2004 and December 2007. Twenty cases were regularly followed up, and the remaining 273 were resected by EMR. We performed univariate and multivariate analyses of clinical and endoscopic characteristics including lesion size, number of biopsy fragments, and endoscopic appearance, in order to analyze the factors affecting histologic discrepancies. Results: Of the 273 lesions resected by EMR, 207 (75.8 %) showed concordant histology, whereas for 51 (18.7 %) the histology was upgraded after endoscopic resection. Lesion size, absence of whitish discoloration, and the presence of spontaneous bleeding were found by univariate analysis to be statistically significant factors predicting an upgraded histology after EMR (P = 0.026, P < 0.001, and P = 0.025, respectively). Multivariate analysis also showed absence of whitish discoloration to be a statistically significant factor influencing histologic discrepancies (P = 0.001, odds ratio 5.29, 95 % confidence interval 1.95 – 14.37). Perforation and bleeding rates associated with EMR for LGD were 0.7 % and 6.2 %, respectively. Twenty patients who did not undergo EMR were followed up for a mean of 22 months, and 3 were revealed to have adenocarcinoma and 1 high-grade dysplasia on the latest histologic exam. Conclusions: We should consider endoscopic resection for gastric LGD that are 2 cm or more in size and do not have whitish discoloration.