Between January 1985 and September 1987, we performed a prospective comparative study between segmental-pancreas transplantation with duct obstruction by neoprene (n = 17) and pancreaticoduodenal transplantation with enteric diversion to a Roux-en-Y intestinal loop (n = 14). All recipients had insulin-dependent diabetes. The immunosuppressive protocol consisted of low doses of the steroids cyclosporin A and azathioprine. Mean follow-up was 16.5 mo for the enteric-diversion group and 13.5 mo for duct-obstructed groups. Two-year patient and pancreas- and kidney-graft actuarial survival rates were 92.9, 75.5, and 74.2%, respectively, in the former group and 92.3, 58.4, and 63.7%, respectively, in the latter group (NS). Five whole-organ grafts were lost (3 vascular thromboses, 1 pancreatitis, 1 rejection), and four segmental grafts were lost (2 vascular thromboses, 1 bleeding, 1 patient's death with functional graft). More surgical complications occurred in the recipients of whole-organ grafts and were often related to the intestinal anastomosis. A satisfactory blood glucose control was observed at 3 mo and 1 yr in both groups. Provocative tests showed higher and prompter insulin secretion in patients with whole-organ grafts. In patients with segmental grafts, the response was lower and delayed with a general tendency to impaired glucose tolerance. A marked hyperinsulinemia after meals was observed in whole-organ graft recipients. Slight nocturnal hyperinsulinemia was observed in both groups. At 1 yr, glycosylated hemoglobin was normal in both groups. The absence of a significant difference between the two groups, in terms of survival and graft function, and the lower surgical complication rate seen with segmental grafts have made us return to neoprene-injected segmental grafts.