A randomized, placebo-controlled trial was performed to assess the effect of magaldrate (800 mg every 4 h) in reducing the rate of upper gastrointestinal tract bleeding among 100 consecutive patients with severe diseases admitted to a general hospital ward. Upper gastrointestinal tract bleeding occurred in 11 of 48 placebo-treated patients and in only 1 of 52 magaldrate-treated patients (p less than 0.01). Endoscopic examination of these patients showed gastric ulcer (two cases), multiple gastric mucosa ulcerations (nine), and no lesions (one). In three patients who received placebo the hemorrhage was clinically relevant and required transfusion of two or more blood units. Patients with two or more risk factors showed a higher rate of gastrointestinal hemorrhage (p less than 0.05). Respiratory failure and treatment with a high dose of corticosteroids were associated with the highest incidence of bleeding (p less than 0.05 for both). The only adverse reaction associated with magaldrate was a mild and self-limiting diarrhea in two cases. We conclude that patients seriously ill admitted to a general hospital ward should be treated with a prophylactic agent against stress-induced ulcer bleeding. Magaldrate is an effective and safe antacid to prevent gastrointestinal tract bleeding in such patients.