Background: Established neuroendocrine signals do not sufficiently account for the exercise-induced increase in glucose production. Using an innovative, yet classical cross-circulation procedure, we studied whether contracting muscle produces a factor that directly stimulates hepatic glycogenolysis. Methods: Isolated rat hindquarters were perfused in series with isolated livers. Results: Stimulation of the sciatic nerve of one or both legs resulted in an increase in force, which rapidly waned. During one-legged contractions, hepatic glucose production increased initially (from −0.9 ± 0.5 (mean ± SE) to 3.3 ± 0.7 µmol/min, p < 0.05). The peak did not differ significantly from that seen after 20 nM of epinephrine (5.1 ± 1.2 µmol/min, p > 0.05). In response to two-legged contractions, the increase in hepatic glucose production (to 5.4 ± 1.3 µmol/min) was higher (p < 0.05) and lasted longer than that seen during one-legged contractions. During contractions, peak hepatic glucose output exceeded concomitant hepatic lactate uptake (p < 0.05), and glucose output decreased to basal levels, while lactate uptake rose to a plateau. Furthermore, in separate experiments an increase in lactate supply to isolated perfused livers increased lactate uptake, but not glucose output. In intact rats, intra-arterial injection of extract made from mixed leg muscle elicited a prolonged increase (p < 0.05) in plasma glucose concentration (from 5.2 ± 0.1 mM to 8.3 ± 1.5 mM). In perfused livers, muscle extract increased glucose output dose dependently. Fractionation by chromatography of the extract showed that the active substance had a MW below 2000. Conclusion: This study provides evidence that contracting skeletal muscle may produce a hormone with a MW below 2000, which enhances hepatic glycogenolysis according to energy needs. Further chemical characterization is warranted.