Segmentary heterotopic autotransplantation of the body and tail of the pancreas, with visceral exocrine drivation and splenosplenic arterio-venous fistulation, was performed in 12 Beagle dogs. An in vivo glucose tolerance test was conducted before the transplantation and 21 days after the operation. There was a significant decrease in the k coefficient (2.884 +/- 0.234 before and 1.878 +/- 0.128 after transplantation) due to reduced peripheral glucose uptake after transplantation (p < 0.001). Overall glucose-stimulated insulin production was decreased after < 0.001). Two populations were identified retrospectively: in 7 dogs insulin response was satisfactory after transplantation (insulin production > or = 50% pretransplantation level) and in 5 the response was below 50%. Glucose tolerance was tested in vitro in the isolated perfused pancreas transplant in 9 dogs, 30 days after the transplantation. Secretory response was assessed according to the early peak of insulin secretion after glucose stimulation. The patterns of insulin secretion were not different before glucose stimulation but became statistically different after stimulation (p < 0.001). In 5 dogs, the response to in vitro glucose stimulation showed an early peak in insulin secretion and in 4 dogs the insulin response came late with no early peak. On histological examination normal (or subnormal) pancreas grafts and grafts in which extensive sclerosis impaired function could be distinguished. There was a significant correlation between the quality of function and the histology, suggesting that containing post-operative sclerosis to a minimum is an important factor in human transplantations.