Two-hundred Wistar rats were allocated to 4 groups. The groups, 3 representing our acute pancreatitis model induced by intrabiliary injection of a trypsin/enterokinase mixture, were studied as follows: (A) no treatment; (B) given a daily 30-ml enema with 20 mg/kg rifaximin; (C) given a daily 30-ml enema with 20 mg/kg rifaximin plus lactitol 0.5 g/kg, and (D) given a daily 30-ml enema with warm saline. A further group of healthy rats was given an intrabiliary injection of 0.15 ml saline. Sacrifices were made after 6,12,24,48 and 72 h of observation. Serial blood samples were drawn to measure pancreatic enzymes and endotoxin. At sacrifice, ascites, lymph nodes, pancreas, spleen, portal vein blood, arterial blood and bile were obtained for bacteriological culture. Both enema treatments brought about a significant improvement in survival. Enema treatments did not affect the serum level of pancreatic enzymes. A time-course increase in endotoxin level was observed in untreated rats. However, significantly decreased levels were observed after both enema treatments. Overall, ascites was the sample most frequently infected. Lymph nodes contiguous to the gut were found to be infected more frequently than those close to major vessels. The histological pancreatic damage was of a significantly lesser degree in both enema treatment groups. Virtually all severe necrotico-hemorrhagic pancreatic lesions were associated with bacterial infection. These data suggest that bacterial translocation plays a relevant role in the outcome of experimental necrotizing pancreatitis. Intra-abdominal spread and lymphatics seem to be the pathways most likely involved in such processes. Colonic cleansing by non-absorbable antibiotics and lactitol seems to exert a beneficial effect on the supervening infection of experimental necrotizing pancreatitis.