In a prospective randomized study, selective intestinal decontamination with norfloxacin was performed during hospitalization in 32 cirrhotic patients with low ascitic fluid total protein levels. The incidence of infections was compared with that in a control group of 31 nontreated cirrhotic patients of similar characteristics. We found a significantly lower incidence of infections [1/32 (3.1%) vs. 13/31 (41.9%); P less than 0.005] and spontaneous bacterial peritonitis [0/32 (0%) vs. 7/31 (22.5%); P less than 0.05] in patients receiving norfloxacin. The lower incidence of extraperitoneal infections [1/32 (3.1%) vs. 7/31 (22.5%); P = 0.052] in the treated group did not reach statistical significance. The incidence of infections [1/28 (3.6%) vs. 9/22 (40.9%); P less than 0.01] and spontaneous bacterial peritonitis [0/28 (0%) vs. 5/22 (22.7%); P less than 0.05] in cirrhotic patients admitted because of ascites was also significantly lower in the treated group. The decrease in the rate of mortality observed in the group undergoing selective intestinal decontamination did not reach statistical significance. These data show that selective intestinal decontamination is useful to prevent spontaneous bacterial peritonitis and extraperitoneal infections in hospitalized cirrhotic patients with low ascitic fluid total protein levels.