We studied prospectively 40 critically ill neurosurgical patients who required prolonged mechanical ventilation to determine the current incidence of stress-related gastroduodenal erosions and ulcers, and to assess endoscopically the efficacy of acid-reducing prophylactic treatment. Nineteen patients were randomized to receive ranitidine plus antacids if necessary to maintain gastric pH at greater than or equal to 4. The remaining 21 patients were given no drug prophylaxis. Gastric pH was significantly (p less than .001) higher in the treated group: 78% of pH readings were at greater than or equal to 4 as compared to 32% in the control group. However, after five study days, incidence and severity of stress lesions were similar in the two groups: nine patients in each group had more than five erosions, one treated patient had a gastric ulcer, and one control patient had duodenal ulcerations. No patient experienced clinically relevant upper GI bleeding. The lack of severe stress bleeding and the low ulcer rate contrast with results from earlier reports on similar patient populations. Furthermore, drug prophylaxis had no detectable benefit, as assessed endoscopically. These findings suggest that routine stress lesion prophylaxis may not be necessary in critically ill patients with comparable risk factors.