We determined the metabolic effects of insulin derived from renal subcapsular islet grafts, either with systemic delivery of insulin through renal venous drainage (REN) or with portal delivery of insulin after renal vein-to-superior mesenteric vein anastomosis (RMA), in streptozotocin-induced diabetic Lewis rats, in comparison with normal rats. After gavage glucose, the plasma glucose responses were similar to normal in REN and RMA rats; however, hyperinsulinemia occurred in REN rats (area under the concentration curves [AUCs] of insulin, 27 +/- 3 nmol x 1(-l) min) in comparison with RMA (14 +/- 2) and normal rats (19 +/- 2), P < 0.003, with no difference in C-peptide responses. The ratio of AUC C-peptide to AUC insulin was lower in REN (2.0 +/- 0.2) than in RMA (3.4 +/- 0.3) and normal animals (3.2 +/- 0.3), P < 0.0005. In euglycemic-hyperinsulinemic clamp studies using the same insulin infusion rate (10 pmol x kg(-1) x min(-1), insulin resistance was found in REN animals (mean glucose infusion rate [GIR], REN: 7.5 +/- 1.2; RMA: 12.0 +/- 1.2; normal: 12.7 +/- 1.0 mg x kg(-1) x min(-1); P < 0.008), with higher steady-state insulin levels in REN (554 +/- 63 pmol/l) than in RMA (291 +/- 26) and normal rats (269 +/- 60), P < 0.0001. With matching steady-state insulin levels in RMA and REN rats during infusion of insulin at 20 pmol x kg(-1) x min(-1) in RMA rats (steady-state insulin 623 +/- 64 pmol/l), GIR was 15.7 +/- 0.7 mg x kg(-1) x min(-1). Thus, systemic delivery of insulin from islet grafts is associated with hyperinsulinemia, insulin resistance, and decreased metabolic clearance of insulin. These abnormalities are prevented by portal delivery of insulin from islet grafts in the same site. The findings are consistent with the hypothesis that portal delivery of insulin is important in maintenance of normal whole-body insulin sensitivity.