SUMMARY Five diabetic patients with renal allografts (placed 2 to 6 years ago) received cadaveric segmental pancreatic grafts (body and tail attached to splenic artery and vein) i.p. with the duct left open and vascular anastomoses to the common iliac vessels. The procedure was performed without technical mishaps in the first three patients who remain well 8 to 10 months later. The fourth patient developed severe cytomegalovirus (CMV) sepsis, had poor graft function, and died. The fifth patient received a graft from a donor with a gastrostomy. Three weeks later peritonitis, secondary to the same organisms grown from the donor gastrostomy site was diagnosed; a functioning graft was removed, but the patient ultimately died of respiratory complications. The three survivors had minimal to no abdominal pain for 2 weeks and none thereafter; mild to no ascites; no fistulas; and no peritonitis, although serum amylase levels were high. All three patients became normoglycemic. Patient 1 remains off insulin with a functioning graft at 10 months after transplantation. Patients 2 and 3 needed no insulin until rejection occurred at 3½ and 2 months. The rejected grafts were not removed, and only resumption of insulin therapy was necessary. This experience shows that the peritoneum can absorb pancreatic secretions and free pancreatic juice can be tolerated in the peritoneal cavity, but there must be no microbial or enteric contamination at operation. Rejection can also be tolerated; the grafts may not need to be removed if there are no other problems. In patient 1, serum amylase has gradually declined, suggesting either that the duct was closing or that exocrine atophy was occurring; however, plasma glucose levels have remained normal. Thus, if rejection or technical complications can be prevented, good endocrine function can be provided by unligated segmental pancreatic grafts Total endocrine replacement therapy for diabetes mellitus is a goal for transplant surgeons, but clinical application of pancreas transplantation has been difficult. Only 3 of 57 pancreas transplants reported to the ACS/NIH Organ Registry functioned for more than 1 year (1). Although rejection was a frequent cause of failure, technical problems related to handling of the pancreatic duct and activation of pancreatic enzymes accounted for much of the morbidity and mortality (2, 3) Pancreaticoduodenal transplantation (3) has largely been abandoned in favor of segmental pancreas (body and tail) transplantation. Various approaches for handling pancreatic duct secretions have included duct ligation (4-7), anastomosis to the recipient ureter (5) or to a Roux-en-Y loop of intestine (7-9), or complete obliteration of the pancreatic ductal system by injection of Neoprene (a synthetic polymer) (10). The pancreatic grafts have been placed in the retroperitoneal iliac fossa of the recipient Pancreatitis and fistula formation have usually occurred after duct ligation (5, 7). Anastomosis of the duct to the host bowel or ureter has been successful, but leaks and fistulas and pancreatic enzyme activation have occurred in some of these cases also (5, 6) Dubernard et al. (10) demonstrated that intraductal injection of Neoprene was a promising technique for selective suppression of exocrine function, obviating the need for duct anastomosis. We evaluated the Neoprene technique of segmental pancreas transplantation in dogs and compared the outcome to recipients of i.p. grafts in which the duct was not ligated and the pancreatic secretions were allowed to drain freely into the peritoneal cavity. We found that there were fewer complications with the duct open than with the Neoprene technique (11). In addition, Kyriakides et al. (12) showed that i.p. transplantation of unligated pancreatic allografts in pigs resulted in few complications After experiments in dogs showed pancreatic juice to be well tolerated in the peritoneal cavity, we adapted this approach for clinical transplantation. Immediately vascularized, segmental pancreas transplants with unligated ducts were placed into the peritoneal cavity of five diabetic patients who had previously received renal allografts for the treatment of end stage renal failure