Dogs with indwelling catheters in the jugular vein and in the carotid artery ran on a treadmill (15% slope, 90 m/min). Hepatic glucose production (Ra), glucose utilization (Rd), and the metabolic clearance rate (MCR) of glucose were measured according to the primed constant rate infusion tracer techniques using 3-3H-glucose as tracer. In addition, plasma insulin (IRK) and glucagon (IRG) levels were determined. In 60 min, exercise increased Ra from 3.8 ± 0.3 to 8.5 ± 0.9 mg/kg min and MCR from 4.0 ± 0.2 to 10.4 ± 0.8 ml/kg min. Plasma insulin declined from 11.4 ± 1.2 to 7.9 ± 0.6 μ-U/ml and glucagon rose from 70 ± 10 to 110 ± 12 pg/ml. After 1 h, after a priming dose (0.36 g/kg), mannoheptulose was infused (6 mg/kg min) for 60 min. IRI fell to 3.4 μU/ml and IRG did not change significantly. Ra rose about 100% (to 16.3 ± 2.1 mg/kg min) and plasma glucose increased from 97 ± 2 to 200 ± 21 mg/dl followed by a slow rise of Rd. The MCR showed a transient decline from 10.3 ± 0.8 to 8.3 ± 1.5 ml/kg min. After that, it did not differ significantly from controls. These effects could be prevented by the infusion of insulin (0.1 mU/kg min). There was a significant inverse correlation between the logarithm of Ra and that of IRI (r = 0.83) and an even stricter correlation between Ra and the logarithm of the molar ratio of IRG/IRI (r = 0.96). It is concluded that the main role of exercise-induced hypoinsulinemia is to increase hepatic glucose production by increasing the sensitivity of the liver to glucagon. Glucose utilization of the working muscle is dominated by the increased activity of hexokinase (s), and physiologic changes of plasma insulin have little effect On it.