These studies were conducted to determine the magnitude and mechanism of compensation for impaired glucagon and insulin responses to exercise. For this purpose, dogs underwent surgery >16 days before experiments, at which time flow probes were implanted and silastic catheters were inserted. During experiments, glucagon and insulin were fixed at basal levels during rest and exercise using a pancreatic clamp with glucose clamped (PC/GC; n = 5), a pancreatic clamp with glucose unclamped (PC; n = 7), or a pancreatic clamp with glucose unclamped + intraportal propranolol and phentolamine hepatic alpha- and beta-adrenergic receptor blockade (PC/HAB; n = 6). Glucose production (R(a)) was measured isotopically. Plasma glucose was constant in PC/GC, but fell from basal to exercise in PC and PC/HAB. R(a) was unchanged with exercise in PC/GC, but was slightly increased during exercise in PC and PC/HAB. Despite minimal increases in epinephrine in PC/GC, epinephrine increased approximately sixfold in PC and PC/HAB during exercise. In summary, during moderate exercise, 1) the increase in R(a) is absent in PC/GC; 2) only a moderate fall in arterial glucose occurs in PC, due to a compensatory increase in R(a); and 3) the increase in R(a) is preserved in PC/HAB. In conclusion, stimulation of R(a) by a mechanism independent of pancreatic hormones and hepatic adrenergic stimulation is a primary defense against overt hypoglycemia.