Intraperitoneal transplantation of encapsulated islets can restore normoglycemia in diabetic recipients but not normal glucose tolerance nor normal insulin responses to a physiological stimulus. This study investigates whether the intraperitoneal implantation site as such contributes to the interference with optimal transport kinetics between the islets and the bloodstream. Insulin was infused into the peritoneal cavity of conscious and freely moving rats in doses of 20, 40, and 80 pmol.l-1.min-1 during 15 min, to mimic the gradual release of insulin from an encapsulated, i.e., a nonvascularized, islet graft. With 20 pmol.l-1.min-1, we observed virtually no rise of insulin levels, and it took 30 min until glucose levels had dropped significantly. With 40 and 80 pmol.l-1.min-1 insulin infusions, there was a dose-dependent rise of insulin and decrease of glucose levels. When compared with intraportal infusions with the same insulin dosages, however, they were strongly delayed and reduced as well as prolonged. Similar results were obtained when inulin instead of insulin was intraperitoneally infused, with indicates that the transport of insulin from the peritoneal cavity to the bloodstream is mainly by passive diffusion. With a view on the clinical efficacy of the bioartificial pancreas, our findings indicate that we should focus on finding or creating a transplantation site that, more than the unmodified peritoneal cavity, permits close contact between the bloodstream and the encapsulated islet tissue.