We randomized 100 critically ill patients at risk of developing acute gastrointestinal ulceration and bleeding into two groups. One (51 patients) received antacid prophylaxis, and the other (49 patients) received no specific form of prophylaxis. Hourly antacid titration kept the pH of the gastric contents above 3.5. Two of the 51 patients who received antacid prophylaxis and gastrointestinal bleeding. Twelve of the 49 control patients bled (P less than 0.005). Of the 12 patients in the control group who bled, seven were placed on antacid medication, and all seven apparently stopped bleeding. Analysis of all the patients showed that an increasing prevalence of respiratory, failure, sepsis, peritonitis, jaundice, renal failure and hypotension was correlated with a greater frequency of bleeding. No patients required operative treatment to control bleeding. These data indicate that the occurrence of acute gastrointestinal bleeding in critically ill patients can be reduced by antacid titration.