The bioartificial pancreas was designed to incorporate islet tissues and a selectively permiable membrane that isolates islets from the immune system of the recipient. The efficacy of agarose, a nontoxic polysaccharide, has been evaluated as a material of microcapsules to prevent allo- and xenograft rejection in rodents. The aim of this study is to demonstrate the possibility of the agarose microcapsule containing allo-islets as a bioartificial pancreas in canine model. In vitro viability of islets was determined by glucose challenge during perifusion experiments (n = 4). Insulin secretion from both encapsulated (enc.) and non-encapsulated (non-enc.) canine islets rose from initial basal levels of 0.09 (encap.), 0.07 (non-encap.) to the peak of 0.2 (encap.), 0.1 (non-encap.) in microU/islet/min after 5 minutes, then decreased to the basal level when the glucose challenge was discontinued. Auto-transplantation was performed in two dogs to evaluated in vivo viability and biocompatibility of encapsulated islets implanted into the splenic sinus by venouse reflux. Two weeks after auto-transplantation, the plasma insulin levels in the splenic vein and artery of two dogs were assayed. In the first dog, serum insulin level was 1 microU/ml both in the vein and the artery and increased, after glucagon (1.0 mg) injection, to levels of 9 microU/ml in the vein, but still kept 1 microU/ml in artery, as well as in the second one. Histological and electron-microscopical examination of the spleen revealed that encapsulated islets remained morphologically intact and the surface of agarose capsules showed no significant adherence of fibroblasts and inflammatory cells. Functional efficacy of the microencapsulated islets was determined using five totally-pancreatectomized diabetic dogs as recipients without immunosuppression. Defined quantity of microencapsulated islets from outbred mongrel donors were grafted through the catheter into omental tissue of the pancreatectomized recipients. All dogs had various degrees of reduced insulin requirements. In three of five recipients, the average fasting glucose values were controlled under 120 mg/dl for 28, 42, 49 days without exogenous insulin, which received totally 4.3 x 10(3), 7.3 x 10(3) and 1.0 x 10(4) (IE/kg) of microencapsulated islets, respectively. In conclusion, the present study indicates that the agarose-based microencapsulated islets can function in large diabetic animals, resulting in the independence of exogenous insulin therapy for prolonged periods without the need for immunosuppression.