Since 1984 the peritoneovenous (LeVeen) shunt has been installed in 33 patients (10 females 30.3%, and 23 males 69.7%), with the average age of 54 +/- 8 years all in the oedema--ascitic decompensation phase of their primary illness. The control group consisted of 39 patients with identical etiology and primary illness stadium, sex and age structure and duration of primary illness, all treated with medicament - diet therapy. All the operated patients and those treated with medicament--diet regime died. Their autopsy findings were confronted. The aim of this study was to indicate the type and frequency of digestive hemorrhage in this population on our own clinical--patient material. Gastrointestinal hemorrhage was the cause of death in 8 (24.24%) patients out of the group operated on, and 6 (15.3%) from the control group. The isolated rupture of esophageal varices was the cause of death in one (3.03%) patient out of the group operated on, and in 5 (12.12%) patients from the control group, the rupture of esophageal varices with toxic vasculitis in 3 (0.09%) patients from the group operated on, and not in one out of the control group; rupture of esophageal varices associated with toxic vasculitis and disorders of the hemostatic mechanism appeared in 4 (12.12%) of patients from the group operated on, and in 1 (2.56%) of the control group. On the basis of the presented it can be concluded that gastrointestinal hemorrhage is a significantly more frequent cause of death in the group operated on than in the control group, and that hemorrhages are of an all the more complex etiology when the flow-in of ascites into circulation is greater. Hemodilution and endotoxins are to blame for their occurrence and it is necessary to preoperatively conduct, beside the control of the hemostatic mechanism, also the test of ones own ascitic burdening and to determine the level of endotoxin in the ascites.