Background: Endoscopic mucosal resection (EMR) is a useful therapeutic technique for colorectal tumors. Yet the maximum lesion diameter to qualify for en bloc resection is about 20 mm due to the physical size limitation of the operating snare. For tumors larger than 20 mm, the necessity of piecemeal resection is higher. If the tumor is resected using piecemeal methods, precise pathological evaluation is difficult. In this case, the risk of residual tumor requires additional operation. In addition, reported local recurrence rates are very high (5-45%). Recently, endoscopic submucosal dissection (ESD) was developed for the resection of early gastric cancer and has been performed for the resection of colorectal tumors in several medical centers. This enables en bloc resection regardless of the size of the tumors. Objective: The aim of this study is to compare the result of EMR, EMR-P (EMR-precutting), and ESD in the treatment of colorectal tumors lager than 20 mm. Methods: We reviewed 254 colorectal tumors without stalk larger than 20 mm which were treated by endoscopic methods (EMR, EMR-P, ESD) from Jan 2004 to Oct 2008. We performed EMR for 92 colorectal tumors, EMR-P for 28 tumors and ESD for 134 tumors. Results: The mean size of tumors was 22.1 ± 3.8 mm (EMR), 21.6 ± 2.9 mm (EMR-P), and 28.2 ± 8.3 mm (ESD). The en bloc resection rates were 28.3% in the EMR group, 57.1% in the EMR-P group, and 88.7% in the ESD group. Perforation during procedure occurred in 2 cases of the EMR-P group and 12 cases of the ESD group. Most of the perforated cases recovered after clipping and only 2 cases of the ESD group received emergency laparoscopic surgery. Delayed bleeding occurred only in two cases of the EMR-P group. The ratio of adenocarcinomas was 10.9% in the EMR group, 28.6% in the EMR-P group, and 42.5% in the ESD group. Within the EMR group, operation was performed on two cases of massive submucosal invasion and four cases of uncertain margin. In the EMR-P group, only one operation was performed due to massive submucosal invasion. In ESD group, nine operations were performed. Among these, there were 6 cases of massive submucosal invasion, 1 case of uncertain margin, and two cases of perforation. Recurrence rates were 34.5% (EMR, median follow-up period: 15 months), 0% (EMR-P: 11 months), and 0% (ESD: 6 months). Conclusion: In the treatment of large colorectal tumors, ESD is more effective than EMR. EMR-P is also very effective but it shares the same operating limitation of EMR. Although ESD is technically demanding, ESD has clinical significance by overcoming the limitation of EMR or EMR-P and facilitating a more precise pathological diagnosis.