It is generally agreed that there is an increased incidence of postoperative renal failure in patients with obstructive jaundice. The proposition that this may be due to endotoxin, and that the endotoxin may be absorbed from the patient's own bowel flora, has been investigated. The study was conducted in patients and in animals. Twenty-four patients with obstructive jaundice were studied. Sixteen had endotoxin in the portal blood at operation and 13 of these had peripheral endotoxaemia at the time of operation or developed it during the 72 hours following operation. Only 3 of the patients had a proved possible site of infective origin for the endotoxaemia, the remainder absorbing the endotoxin from their own gastro-intestinal organisms. Portal endotoxaemia occurred when the serum bilirubin was 8-5 mg/100 ml or greater. There was a highly significant decrease in mean endogenous creatinine clearance postoperatively in patients with endotoxaemia, and expression of this as a percentage postoperative fall compared with preoperative levels enhanced the significance. Jaundiced patients without endotoxaemia had no significant fall in endogenous creatinine clearance. Matched non-jaundiced control patients did not develop portal or peripheral endotoxaemia, and there was no significant fall in postoperative creatinine clearance. Experiments using animals showed that the absence of bile salts in the intestinal tract in obstructive jaundice allows endotoxin to be absorbed, and that this absorption can be prevented by the oral administration of bile salts. The therapeutic implications in patients are discussed.