Efficacy of primary endoscopic hemostasis may be achieved in 87-100% of cases with rate of recurrences of bleeding after successful hemostasis up to 13-35%. Exactly this category of patients are most likely to be operated on - 6.5-21% of patients. 517 patients with peptic ulcer GI bleeding are treated from 2010 to 2012. Surgery was performed in 45 (8.7%) patients. By comparison of groups of patients who were operated and treated nonsurgically test parameters were patients age, localization and condition of ulcerous defects, co-morbidities, signs of hemorrhagic shock at admission and recurrent hemorrhage, chosen kinds of endoscopic hemostasis. As a systematized prognostic method Rockall and Glasgow-Blatchford scores were used. Signs of shock at admission were revealed in 34 (73.9%) patients who underwent surgery and 90 (19%) patients of nonsurgically treated group of patients (p=0.001). Severity of hemorrhage was estimated by number of concentrated red cells (450 ml) transfused: 5.3±3.7 units in group of surgically treated patients and 0.9±2.8 units in group of patients treated conservatively (p=0.001). Source of bleeding in the 1st and 2nd groups in 48.9% and 57% of patients respectively were duodenal ulcers. Ulcerous defects larger than 2 cm were detected 22 (48.9%) operated patients and 63 (13.5%) conservatively treated patients (p=0.001). Recurrence of bleeding developed in 37 (82.2%) operated patients and 15 (3.2%; p=0.001) conservatively treated patients. Overall mortality comprised 8.7%. Predisposing to surgical treatment criteria are long-lasting ulcerous past medical history, signs of hemorrhagic shock at admission, inpatient bleeding event, ulcerous defect larger than 2 cm, active bleeding or visible vessel in the bottom of the ulcer (FIa-FIIa) and penetration of the ulcer. Likelihood of surgical treatment as well as death may be predicted with the use Rockall and Glasgow-Blatchford scores.