The success of the extravascular bioartificial pancreas (BAP) is contingent on the rapid transfer of the glycemic signal across both an extravascular compartment and a semipermeable membrane and of insulin from the BAP to the recipient. To examine the possibility of microencapsulated islets such as the BAP to achieve satisfactory in vivo glucose-insulin kinetics, islets were isolated from Lewis rats, encapsulated in poly-L-lysine-alginate membranes, and isogenically transplanted into the peritoneal cavity of 14 streptozotocin induced diabetic rats. Fasting blood glucose (BG) was measured and intravenous glucose tolerance was tested at 8-10 weeks and compared with three control groups: 1) normal Lewis rats (n = 6); 2) untreated diabetic rats (n = 5); and 3) diabetic rats that received intraperitoneal implants of empty capsules (n = 4). Ten animals that received microencapsulated islets (5,271 +/- 431) promptly became normoglycemic, with a mean BG of 128 +/- 17 ng/dl 3 days after transplantation and maintained this level > 100 days. Intravenous glucose tolerance K-value for the group was 3.84 +/- 0.32 compared with 3.96 +/- 0.39 (p = 0.83) for the normal control group, and 0.60 +/- 0.12 (p < 0.01) and 0.40 +/- 0.15 (p < 0.01) for the diabetic control groups with and without empty capsules. The authors conclude from these results that, given sufficient beta-cell mass, a BAP without any vascular access can respond appropriately to an increase in blood glucose concentration, without overshoot hypoglycemia and within a lag lapse compatible with normal physiologic insulin delivery.(ABSTRACT TRUNCATED AT 250 WORDS)