Acute upper gastrointestinal (GI) bleeding is an actual problem of modern medicine. Prediction of recurrent bleeding after successful treatment is one of the main issues. There are various prognostic scales for the prognosis of recurrent bleeding. Objective — to optimize diagnosis and treatment of patients with acute upper GI bleeding via validity analysis and clinical introduction of the integrated prognostic scales Rockall and Glasgow—Blatchford in a modern multi-field hospital. Material and methods. The most reliable international scales Rockall and Glasgow—Blatchford were used because there is no implemented unified prognostic scale in Russia for the prediction of recurrent bleeding. A prospective study included 120 patients who were hospitalized with acute non-varicose upper GI bleeding. Patients were divided into groups of low (0—2 scores), medium (3—7 scores) and high risk (8—12 scores) in accordance with the Rockall scale and into groups of low (0—5 scores) and high risk (6—24 scores) in accordance with the Glasgow—Blatchford scale. Prognosis was calculated using these scores and compared with actual clinical outcomes. These data were studied using logistic regression and ROC analysis. We assessed sensitivity, specificity of the method and validity of the scales. Results. Patients with low scores on both scales are characterized by the minimal likelihood of recurrent bleeding. Increased risk of recurrent bleeding is observed in patients with Rockall score 8—12 and Glasgow—Blatchford score 6—24. Conclusion. High validity of the Rockall and Glasgow — Blatchford scales has been proven in predicting recurrent upper gastrointestinal bleeding. Rockall scale is characterized by a more reliable prognosis.