In an attempt to determine the optimal duration of therapy of spontaneous bacterial peritonitis, 100 patients with neutrocytic ascites and suspected spontaneous bacterial peritonitis were randomized to short-course vs. long-course treatment groups. Empiric therapy was initiated before the results of ascitic fluid culture were available. Of the 90 patients who met strict criteria for spontaneous bacterial peritonitis or culture-negative neutrocytic ascites, 43 were randomized to a group receiving 5 days and 47 to a group receiving 10 days of single-agent cefotaxime, 2 g IV every 8 hours. Infection-related mortality (0% vs. 4.3%), hospitalization mortality (32.6% vs. 42.5%), bacteriologic cure (93.1% vs. 91.2%), and recurrence of ascitic fluid infection (11.6% vs. 12.8%) were not significantly different between the 5- and 10-day treatment groups, respectively. Recurrence rates were comparable to the values reported in the literature. The cost of antibiotic and antibiotic administration were significantly lower in the short-course group. Short-course treatment of spontaneous bacterial peritonitis is as efficacious as long-course therapy and significantly less expensive.