Aim: To assess how many patients with benign "difficult" colorectal lesions (DCRLs) referred to surgical resection, may be treated with endoscopic resection (ER) rather than surgical resection., Methods: The prospectively collected colonoscopy database of our Endoscopic Unit was reviewed to identify all consecutive patients who, between July 2011 and August 2013, underwent an endoscopic re-evaluation before surgical resection due to the presence of DCRLs with a histological confirmation of benignancy on forceps biopsy. ER was attempted when the lesion did not have definite features of deeply invasive cancer. The "nonlifting sign" excluded ER only in naive lesions without a prior attempted resection. Lesions were classified, using the Kyoto-Paris classification for mucosal neoplasia. For sessile and non-polypoid lesions the "inject and cut" resection technique was used. Pedunculated and semi-pedunculated lesions were transected at the stalk just below the polyps head and before or after resection, metal clips or a loop were applied on the stalk to prevent bleeding. The lesions were histologically classified according to the Vienna criteria and for the pedunculated lesions the Haggitt classification was used., Results: Eighty-two patients (42 females, mean age 62 years) with 82 lesions (mean size 37 mm) were included in the study. Sixty-nine (84%) lesions were endoscopically resected, while 13 underwent surgical resection since ER was deemed unsuitable. On histology, cancer was found in 21/69 lesions (14 intra-mucosal, 7 sub-mucosal) and was associated with the size (P < 0.001) and with type 0-IIa +Is (P = 0.011) and 0-IIa + IIc (P < 0.001) lesions. All patients with sub-mucosal cancer, underwent surgical resection. Complications occurred in 11/69 patients (7 bleedings, 2 transmural burn syndromes, 2 perforations), all managed endoscopically or conservatively, and were associated with presence of invasive cancer (P = 0.021). During follow-up recurrence/residual tissue was found in 14/51 sessile or non-polypoid lesions (13 treated endoscopically, 1 underwent surgical resection) and was associated with type 0-IIa + Is lesions (P = 0.001), piecemeal resections (P = 0.01) and with lesion size (P = 0.004). Overall, 74% of patients avoided surgery. Surgical resection was significantly associated with type 0-IIa + Is (P = 0.01) and 0-IIa + IIc (P = 0.001) lesions, with sub-mucosal invasion on histology (P < 0.001), with presence of the "nonlifting sign" (P < 0.001), and related to the dimension of the lesions (P = 0.001). In the logistic regression analysis, the only independent predictor for surgical resection was the dimension of the lesions (P = 0.002)., Conclusion: Before submitting patients to surgical resection for a benign DCRL, a second opinion by an experienced endoscopist is mandatory to avoid unnecessary surgery.