This randomized prospective study was aimed at assessing the efficiency of a systemic antibiotic therapy for the prevention of bacterial infections in cirrhotic patients with gastrointestinal hemorrhage by ruptured esophageal varices. For 15 mo, all patients hospitalized with no infection on admission, were included in the study. Starting on admission day, patients in group A received ofloxacin (400 mg/day) for 10 days, first intravenously then orally. They also received an intravenous bolus of amoxicillin plus clavulanic acid (1 g) before each endoscopy performed during hemorrhage. Patients in group B received antibiotic therapy only in cases of established or suspected infection. Chest X-ray, blood culture, urine culture and sputum and ascitic fluid culture were performed every day for 7 days, then every other day for the next 7 days. A bronchial sampling was performed with the Wimberley technique on patients with endotracheal intubation. Ninety-one patients (55 men, 54 +/- 11 years, 78% Child Pugh class C) were included in the study (46 in group A, 45 in group B). Group A showed a lower incidence of bacterial infections than group B (20% vs. 66%; p < 0.001). Breakdown of positive bacteriological sampling was as follows: blood (6 vs. 17), ascites (3 vs. 7), lungs (2 vs. 18), urine (1 vs. 10). The 2-wk mortality rate was 24% in group A and 35% in group B.