The effects of intragastric prostaglandin E2 (PGE2) on the occurrence of acute GI hemorrhage in intensive care patients were investigated in a prospective, double-blind, placebo-controlled study. Ninety patients with two or more risk factors (major surgery, multiple trauma, respiratory insufficiency, renal insufficiency, jaundice, hypotension, peritonitis, sepsis) were randomized for treatment with either PGE2 (0.5 mg) or placebo, administered every 4 h via a nasogastric tube. Blood loss in gastric aspirates was measured by 51Cr-erythrocyte labeling and a peroxidase test (orthotolidine). Of 57 patients who could be evaluated, 29 received PGE2 and 28 received placebo. Hemorrhage occurred in nine PGE2 patients and 13 placebo-treated patients, not a significant difference. The occurrence of hemorrhage was related to the number of risk factors, and GI hemorrhage was rarely the major factor determining mortality. Results of the orthotolidine test were not positively correlated with those of erythrocyte labeling, indicating that peroxidase tests should not be relied upon to detect blood in gastric aspirates.