We assessed basal glucose metabolism in 16 female nonpregnant (NP) and 16 late-pregnant (P) conscious, 18-h-fasted dogs that had catheters inserted into the hepatic and portal veins and femoral artery approximately 17 days before the experiment. Pregnancy resulted in lower arterial plasma insulin (11 +/- 1 and 4 +/- 1 microU/ml in NP and P, respectively, P < 0.05), but plasma glucose (5.9 +/- 0.1 and 5.6 +/- 0.1 mg/dl in NP and P, respectively) and glucagon (39 +/- 3 and 36 +/- 2 pg/ml in NP and P, respectively) were not different. Net hepatic glucose output was greater in pregnancy (42.1 +/- 3.1 and 56.7 +/- 4.0 micromol. 100 g liver(-1).min(-1) in NP and P, respectively, P < 0.05). Total net hepatic gluconeogenic substrate uptake (lactate, alanine, glycerol, and amino acids), a close estimate of the gluconeogenic rate, was not different between the groups (20.6 +/- 2.8 and 21.2 +/- 1.8 micromol. 100 g liver(-1). min(-1) in NP and P, respectively), indicating that the increment in net hepatic glucose output resulted from an increase in the contribution of glycogenolytically derived glucose. However, total glycogenolysis was not altered in pregnancy. Ketogenesis was enhanced nearly threefold by pregnancy (6.9 +/- 1.2 and 18.2 +/- 3.4 micromol. 100 g liver(-1).min(-1) in NP and P, respectively), despite equivalent net hepatic nonesterified fatty acid uptake. Thus late pregnancy in the dog is not accompanied by changes in the absolute rates of gluconeogenesis or glycogenolysis. Rather, repartitioning of the glucose released from glycogen is responsible for the increase in hepatic glucose production.