We have previously shown that during glucose clamps in moderately hyperglycemic depancreatized dogs: 1) peripheral insulin infusion, resulting in greater systemic insulinemia and greater suppression of glucagon than equidose portal infusion, inhibited glucose production (GP) to a greater extent; and 2) portal and half-dose peripheral infusions, resulting in matched peripheral insulinemia and similar suppression of glucagon, inhibited GP equally. These findings are consistent with an indirect effect of insulin in suppressing GP in diabetic dogs, which might be partly mediated by the differential suppression of glucagon. To address this question, we performed the experimental protocols of the previous study under conditions of constant glucagon levels (approximately 550 ng/liter), achieved by a high rate portal glucagon infusion (5 ng/kg.min). As in the previous study (basal glucagon levels, approximately 170 ng/liter), we used depancreatized dogs and assessed GP with HPLC-purified [6(-3)H]glucose. After obtaining constant basal hyperglycemia (approximately 10 mM) with portal infusions of insulin (4.8 +/- 0.5 pmol/kg.min) and glucagon, an additional infusion of insulin was administered for 180 min, either portally (portal; n = 7) or peripherally (peripheral; n = 8) at the same rate (5.4 pmol/kg.min) or at half that rate peripherally (1/2 periph; n = 5). Plasma glucose and glucose specific activities were clamped at basal levels. Systemic insulin levels increased by 215 +/- 16,310 +/- 26, and 184 +/- 15 pM, and estimated hepatic insulin levels increased by 398 +/- 20, 310 +/- 26, and 184 +/- 15 pM with portal, peripheral, and 1/2 periph, respectively. GP was suppressed to the same extent with portal and peripheral (53 +/- 6% and 50 +/- 6%), but less with 1/2 periph (35 +/- 5%). FFA levels were suppressed to a greater extent with peripheral than portal or 1/2 periph, whereas the responses of lactate alanine and glycerol to insulin infusion were similar in the three groups. Thus, in the present report, unlike in our previous study, 1) suppression of GP was proportional to the hepatic insulin levels; and 2) systemic insulin levels did not dominate suppression of GP. We, therefore, conclude that in hyperglycemic depancreatized dogs 1) glucagon, at concentrations seen in poorly controlled diabetes, can unmask a direct effect of hepatic insulin levels on GP; and 2) the suppression of glucagon may play a role in the peripheral effect of exogenously delivered insulin on GP. This is the first in vivo study to show that the main direct effect of insulin on the liver is to counteract the effect of glucagon.