After a meal, glucagon-like peptide-1 (GLP-1) levels in the hepatic portal vein are elevated and are twice those in peripheral blood. The aim of this study was to determine whether any of GLP-1's acute metabolic effects are initiated within the hepatic portal vein. Experiments consisted of a 40-min basal period, followed by a 240-min experimental period, during which conscious 42-h-fasted dogs received glucose intraportally (4 mgxkg(-1)xmin(-1)) and peripherally (as needed) to maintain arterial plasma glucose levels at approximately 160 mg/dl. In addition, saline was given intraportally (CON; n = 8) or GLP-1 (1 pmolxkg(-1)xmin(-1)) was given into the hepatic portal vein (POR; n = 11) or the hepatic artery (HAT; n = 8). Portal vein plasma GLP-1 levels were basal in CON, 20x basal in POR, and 10x basal in HAT, whereas levels in the periphery and liver were the same in HAT and CON. The glucose infusion rate required to maintain hyperglycemia was significantly greater in POR (8.5 +/- 0.7 mgxkg(-1)xmin(-1), final 2 h) than in either CON or HAT (6.0 +/- 0.5 or 6.7 +/- 1.0 mgxkg(-1)xmin(-1), respectively). There were no differences among groups in either arterial plasma insulin (24 +/- 2, 23 +/- 3, and 23 +/- 3 microU/ml for CON, POR, and HAT, respectively) or glucagon (23 +/- 2, 30 +/- 3, and 25 +/- 2 pg/ml) levels during the experimental period. The increased need for glucose infusion reflected greater nonhepatic as opposed to liver glucose uptake. GLP-1 infusion increased glucose disposal independently of changes in pancreatic hormone secretion but only when the peptide was delivered intraportally.