AIM: To analyze the importance in predicting patients risk of mortality due to upper gastrointestinal (UGI) bleeding under today’s therapeutic regimen. METHODS: From 1998 to 2001, 121 patients with the diagnosis of UGI bleeding were treated in our hospital. Based on the patients’ data, a retrospective multivariate data analysis with initially more than 270 single factors was performed. Subsequently, the following potential risk factors underwent a logistic regression analysis: age, gender, initial hemoglobin, coumarines, liver cirrhosis, prothrombin time (PT), gastric ulcer (small curvature), duodenal ulcer (bulbus back wall), Forrest classification, vascular stump, variceal bleeding, Mallory-Weiss syndrome, RBC substitution, recurrent bleeding, conservative and surgical therapy. RESULTS: Seventy male (58%) and 51 female (42%) patients with a median age of 70 (range: 21-96) years were treated. Their in-hospital mortality was 14%. While 12% (11/91) of the patients died after conservative therapy, 20% (6/30) died after undergoing surgical therapy. UGI bleeding occurred due to duodenal ulcer (n = 36; 30%), gastric ulcer (n = 35; 29%), esophageal varicosis (n = 12; 10%), Mallory-Weiss syndrome (n = 8; 7%), erosive lesions of the mucosa (n = 20; 17%), cancer (n = 5; 4%), coagulopathy (n = 4; 3%), lymphoma (n = 2; 2%), benign tumor (n = 2; 2%) and unknown reason (n = 1; 1%). A logistic regression analysis of all aforementioned factors revealed that liver cirrhosis and duodenal ulcer (bulbus back wall) were associated risk factors for a fatal course after UGI bleeding. Prior to endoscopy, only liver cirrhosis was an assessable risk factor. Thereafter, liver cirrhosis, the location of a bleeding ulcer (bulbus back wall) and patients’ gender (male) were of prognostic importance for the clinical outcome (mortality) of patients with a bleeding ulcer. CONCLUSION: Most prognostic parameters used in clinical routine today are not reliable enough in predicting a patient’s vital threat posed by an UGI bleeding. Liver cirrhosis, on the other hand, is significantly more frequently associated with an increased risk to die after bleeding of an ulcer located at the posterior duodenal wall. METHODS: From 1998 to 2001, 121 patients with the diagnosis of UGI bleeding were treated in our hospital. Based on the patients The following potential risk factors underwent a logistic regression analysis: age, gender, initial hemoglobin, coumarines, liver cirrhosis, prothrombin time (PT), gastric RESULTS: Seventy male (58%) and 51 female (58%) and ulcer (small curvature), duodenal ulcer (bulbus back wall), Forrest classification, vascular stump, variceal bleeding, Mallory- Weiss syndrome, RBC substitution, recurrent bleeding, Their in-hospital mortality was 14%. While 12% (11/91) of the patients died after conservativ (42%) (n = 36; 30%), gastric ulcer (n = 35; 29%), esophageal varicosis (n = 12; 10%), Mallory-Weiss syndrome (n = 8; 7%), erosive lesions of the mucosa (n = 20; 17% A logistic regression analysis of all of the causes of liver cirrhosis and duodenal ulcer (bulbus) (n = 2; 2%), benign tumor Prior to endoscopy, only liver cirrhosis was an assessable risk factor. After, liver cirrhosis, the location of a bleeding ulcer (bulbus back wall) and patients’ gender (male ) of of prognostic importance for the clinical outcome (mortality) of patients with a bleeding ulcer. CONCLUSION: Most prognostic parameters used in clinical routine today are not reliable enough in predicting a patient’s vital threat posed by an UGI bleeding. Liver cirrhosis, on the other hand, is more more frequently associated with an increased risk to die after bleeding of an ulcer located at the posterior duodenal wall.