Eighty cirrhotic patients who had recovered from an episode of spontaneous bacterial peritonitis were included in a multicenter, double-blind trial aimed at comparing long-term norfloxacin administration (400 mglday; 40 patients) vs. placebo (40 patients) in the prevention of spontaneous bacterial peritonitis recurrence. At entry, both groups were similar with respect to clinical and laboratory data, ascitic fluid protein and polymorphonuclear concentrations, number of previous episodes of spontaneous bacterial peritonitis and causative organisms of the index spontaneous bacterial peritonitis. Norfloxacin administration produced a selective intestinal decontamination (elimination of aerobic gram-negative bacilli from the fecal flora without significant changes in other microorganisms) throughout the study in six patients in whom the effect of norfloxacin on the fecal flora was periodically assessed. Fourteen patients from the placebo group (35%) and five from the norfloxacin group (12%) developed spontaneous bacterial peritonitis recurrence during follow-up ($ = 5.97; p = 0.014) (mean follow-up period = 6.4 f 0.6 mo; range = 1 to 19 mo). Ten of the 14 spontaneous bacterial peritonitis recurrences in the placebo group and only one of the five spontaneous bacterial peritonitis recurrences in the norfloxacin group were caused by aerobic gramnegative bacilli ($ = 8.87; p = 0.0029). The overall probability of spontaneous bacterial peritonitis recurrence at 1 yr of follow-up was 20% in the norfloxacin group and 68% in the placebo group (p = 0.0063) and the probability of spontaneous bacterial peritonitis recurrence caused by aerobic gram-negative bacilli at 1 yr of follow-up was 3% and 60%, respectively (p = 0.0013). Only one patient treated with norfloxacin experienced side effects related to treatment (oral and