OBJECTIVE: To assess three anti-stress ulcer prophylaxis regimens in mechanically ventilated patients for bacterial colonization, early- and late-onset nosocomial pneumonia, and gastrointestinal bleeding. DESIGN: Randomized controlled trial. PATIENTS: Consecutive eligible patients with mechanical ventilation and a nasogastric tube. Of 258 eligible patients, 244 were assessable. SETTING: Medical and surgical intensive care units. INTERVENTION: At intubation, patients were randomly assigned to receive one of the following: antacid (a suspension of aluminum hydroxide and magnesium hydroxide), 20 mL every 2 hours; ranitidine, 150 mg as a continuous intravenous infusion; or sucralfate, 1 g every 4 hours. MEASUREMENTS: Using predetermined criteria, the incidence of gastric bleeding, gastric colonization, early-onset pneumonia, and late-onset pneumonia was assessed in patients intubated for more than 24 hours. RESULTS: Of 244 assessable patients, macroscopic gastric bleeding was observed in 10%, 4%, and 6% of patients assigned to receive sucralfate, antacid, and ranitidine, respectively (P > 0.2). The incidence of early-onset pneumonia was not statistically different among the three treatment groups (P > 0.2). Among the 213 patients observed for more than 4 days, late-onset pneumonia was observed in 5% of the patients who received sucralfate compared with 16% and 21% of the patients who received antacid or ranitidine, respectively (P = 0.022). Mortality was not statistically different among the three treatment groups. Patients who received sucralfate had a lower median gastric pH (P < 0.001) and less frequent gastric colonization compared with the other groups (P = 0.015). Using molecular typing, 84% of the patients with late-onset gram-negative bacillary pneumonia were found to have gastric colonization with the same bacteria before pneumonia developed. CONCLUSION: Stress ulcer prophylaxis with sucralfate reduces the risk for late-onset pneumonia in ventilated patients compared with antacid or ranitidine.