Insulin may suppress hepatic glucose production directly, or indirectly via suppression of release of gluconeogenic substrates from extrasplanchnic tissues. To compare these mechanisms, we performed insulin dose-response experiments in conscious dogs at euglycemia, during somatostatin infusion, and intraportal glucagon replacement. Insulin was sequentially infused either intraportally (0.05, 0.20, 0.40, 1.0, 1.4, and/or 3.0; protocol I) or systemically at half the intraportal rate (0.025, 0.10, 0.20, 0.50, 0.70, and/or 1.5 mU.min-1.kg-1; protocol II). Exogenous glucose infused during clamps was labeled with 3-[3H]glucose (2 microCi/g) to prevent a fall in plasma specific activity (P greater than 0.2) that may have contributed to previous underestimations of hepatic glucose output (HGO). Portal insulins were up to threefold higher during intraportal infusion, but peripheral insulin levels were not different between the intraportal and systemic protocols [7 +/- 5 vs. 9 +/- 1, 12 +/- 4 vs. 13 +/- 6, 16 +/- 3 vs. 27 +/- 5, 70 +/- 23 vs. 48 +/- 8, 83 +/- 3 vs. 86 +/- 21, and 128 vs. 120 +/- 14 microU/ml for paired insulin doses; P greater than 0.06 by analysis of variance (ANOVA)]. Despite higher portal insulin levels in protocol I, HGO suppression was equivalent in the two protocols when systemic insulin was matched, from 3.3 +/- 0.1 to near-total suppression at 0.3 mg.min-1.kg-1 at the highest insulin infusion rate (3.0 mU.min-1.kg-1; P less than 0.0001) with intraportal insulin, from 2.9 +/- 0.8 to -0.8 +/- 0.2 mg.min-1.kg-1 in protocol II (P less than 0.001). Suppression of HGO was similar at matched systemic insulin, regardless of portal insulin, suggesting the primacy of insulin's action on the periphery in its restraint of hepatic glucose production.