Pancreatic infection is the main indication for surgery and the principal determinant of prognosis in acute necrotizing pancreatitis. Previous studies on the effects of antibiotics have not, however, uniformly demonstrated any reduction in the need for surgery or any decrease in mortality among these patients, although the incidence of pancreatic infections was significantly reduced. This single-center randomized study was designed to compare early vs. delayed imipenem treatment for acute necrotizing pancreatitis. Ninety patients with acute necrotizing pancreatitis (C-reactive protein > 150 mg/L, necrosis on CT) were randomized within 48 hours either to a group receiving imipenem (1.0 g plus cilastatin intravenously 3 times a day) or a control group. Not included were those who had been started on antibiotics at the referring clinic, those who were taken directly to the intensive care unit for multiorgan failure, and those who refused antibiotics or might have had adverse reactions. Thirty-two patients were excluded because they were over 70 years of age (not potentionally operable) or for any study violation. There were 25 patients in the imipenem group and 33 patients in the control group. The main end point was the indication for necrosectomy due to infection (i.e., after the initial increase and decrease, there was a second continuous increase in temperature, white blood cell count [> 30%] and C-reactive protein [> 30%], with other infections ruled out, or bacteria were found on Gram stain of the pancreatic fine-needle aspirate). In the control group, imipenem was started when the operative indication was fulfilled. Conservative treatment was continued for at least 5 days before necrosectomy. The study groups did not differ from each other with regard to sex distribution, patient age, etiology, C-reactive protein concentration, and extent of pancreatic necrosis on CT. Two (8%) of 25 patients in the imipenem group compared to 14 (42%) of 33 in the control group fulfilled the operative indications (P = 0.003). Nine patients in the control group responded to delayed antibiotics but five had to undergo surgery. Of those receiving antibiotics, 2 (8%) of 25 in the early antibiotic (imipenem) group needed surgery compared to 5 (36%) of 14 in the delayed antibiotic (control) group (P = 0.04). Two (8%) of 25 patients in the imipenem group and 5 (15%) of 13 patients in the control group died (P = NS [no significant difference]). Seven (28%) of 25 in the imipenem group and 25 (76%) of 33 in the control group had major organ complications (P = 0.0003). Based on the preceding criteria, early imipenem-cilastatin therapy appears to significantly reduce the need for surgery and the overall number of major organ complications in acute necrotizing pancreatitis, and reduces by half the mortality rate; this is not, however, statistically significant in a series of this size.