To examine the relationship between the magnitude of the negative arterial-portal glucose gradient and net hepatic glucose uptake, two groups of 42-h fasted, conscious dogs were infused with somatostatin, to suppress endogenous insulin and glucagon secretion, and the hormones were replaced intraportally to create hyperinsulinemia (3- to 4-fold basal) and basal glucagon levels. The hepatic glucose load to the liver was doubled and different negative arterial-portal glucose gradients were established by altering the ratio between portal and peripheral vein glucose infusions. In protocol 1 (n = 6) net hepatic glucose uptake was 42.2 +/- 6.7, 35.0 +/- 3.9, and 33.3 +/- 4.4 mumol.kg-1.min-1 at arterial-portal plasma glucose gradients of -4.1 +/- 0.9, -1.8 +/- 0.4, and -0.8 +/- 0.1 mM, respectively. In protocol 2 (n = 6) net hepatic glucose uptake was 26.1 +/- 2.8 and 12.2 +/- 1.7 mumol.kg-1.min-1 at arterial-portal plasma glucose gradients of -0.9 +/- 0.2 and -0.4 +/- 0.1 mM, respectively. No changes in the hepatic insulin or glucose loads were evident within a given protocol. Although net hepatic glucose uptake was lower in protocol 2 when compared with protocol 1 (26.1 +/- 2.8 vs. 33.3 +/- 4.4 mumol.kg-1.min-1) in the presence of a similar arterial-portal plasma glucose gradient (-0.9 vs. -0.8 mM) the difference could be attributed to the hepatic glucose load being lower in protocol 2 (i.e., hepatic fractional glucose extraction was not significantly different) primarily as a result of lower hepatic blood flow. In conclusion, in the presence of fixed hepatic glucose and insulin loads, the magnitude of the negative arterial-portal glucose gradient can modify net hepatic glucose uptake in vivo.