The mechanism of vagal nerve stimulation of glucagon (IRG) and insulin (IRI) secretion was investigated in halothane-anesthetized dogs. Both ventral and dorsal branches of the thoracic vagi were stimulated electrically (10 Hz, 5 msec, 13.5 mA, 10 min) below the heart. Arterial and superior pancreaticoduodenal venous plasma were sampled, superior pancreaticoduodenal venous plasma flow was measured, and net pancreatic output of IRG and IRI were calculated. During vagal nerve stimulation (n = 15) net pancreatic output of IRG doubled (delta = +0.83 +/- 0.28 ng/min, P less than 0.01; baseline = 0.81 +/- 0.15 ng/ min) and IRI quadrupled (delta = +3.5 +/- 1.5 mU/min, P less than 0.025; baseline = 1.1 +/- 0.3 mU/min). Arterial glucose levels increased by 7 +/- 2 mg/dl from 108 +/- 3 mg/dl (P less than 0.005). After atropine pretreatment (n = 7), the pancreatic IRI response to vagal nerve stimulation was +0.71 +/- 0.28 mU/min (P less than 0.025), a reduction of 80%. In contrast, atropine pretreatment changed neither the IRG response (delta = +0.87 +/- 0.36 ng/min; P less than 0.05) nor the arterial glucose response (delta = +9 +/- 3 mg/dl; P less than 0.025) to vagal nerve stimulation. Hexamethonium pretreatment (n = 9) abolished the pancreatic IRG response (delta = +0.13 +/- 0.11 ng/min; NS), the arterial glucose response (delta = +0.5 +/- 1.9 mg/dl; NS) and the pancreatic IRI response (delta = +0.16 +/- 0.31 mU/min; NS) to vagal nerve stimulation. It is concluded that vagal nerve stimulation in the dog produces a moderate increase of IRG secretion, mediated by a nonmuscarinic (peptidergic?) mechanism, and a marked increase of IRI secretion, mediated by a muscarinic mechanism. Both responses are dependent on nicotinic transmission.