We infused glucagon into normal humans while preventing changes in plasma glucose and insulin. Insulin (0.45 mU . min-1 . kg-1) were infused for 90 min, while euglycemia was maintained by a variable glucose infusion. Subsequently, glucagon (6 ng . min-1 . kg-1) was added, and changes in plasma glucose were avoided by appropriately reducing the glucose infusion. With insulin alone, glucose production (GP) fell to zero. When hyperglucagonemia (530 +/- 32 pg/ml) was superimposed, GP rose promptly and then slowly declined. However, between 180 and 240 min, GP remained elevated (1.72 +/- 0.30 mg . min-1 . kg-1) as compared to an insulin control study (0.03 +/- 0.20, P less than 0.025). When hyperglycemia (+25 mg/100 ml) was induced between 180 and 240 min, glucagon-stimulated GP was completely suppressed. To determine whether this effect was mediated by hyperglycemia per se or glucose-induced hyperinsulinemia, between 180 and 240 min we increased either a) the insulin infusion (by 0.25 mU . min-1 . kg-1) while maintaining euglycemia or b) plasma glucose (+25 mg/100 ml) while blocking insulin release with somatostatin. When the insulin was increased, GP declined by 68 +/- 13% (P less than 0.02). When plasma glucose alone was raised, GP fell from 1.44 +/- 0.09 to 0.07 +/- 0.16 mg . min-1 . kg-1 (less than 0.002). In conclusion, the hepatic response to sustained hyperglucagonemia is more persistent if changes in plasma glucose are prevented, and its transient nature is in part explained by a feedback adjustment to glucagon-induced hyperglycemia and hyperinsulinemia.