The effect of selective vagotomy of the abomasum, pylorus, duodenum and liver on insulin release during the cephalic phase of digestion was investigated in wethers and lactating ewes. Electrical stimulation of the cervical vagus nerves was carried out to test the completeness of the vagotomies performed. In experiment 1, using wethers, the abomasal, pyloric and duodenal branches (ADV; n = 7) or the hepatic, abomasal, pyloric and duodenal branches (HADV; n = 10) of the ventral and/or dorsal vagus nerves were cut; a third group of wethers underwent sham-operation (SO; n = 8). In experiment 2, vagotomy (ADV; n = 5) or sham-operations (SO; n = 5) were carried out in lactating ewes. Jugular blood was drawn before and after presentation of food for glucose and insulin determination (experiments 1 and 2) or before, during and after the electrical stimulation of the peripheral ends of the cut cervical vagus nerves in randomly selected lactating ewes (experiment 3: ADV = 3, SO = 3) and wethers (experiment 4: ADV = 4, HADV = 4, SO = 4), for determination of insulin only. Presentation of food caused an immediate and significant (P < 0.05) rise in plasma insulin levels in SO animals compared with ADV or HADV wethers (experiment 1) or ADV ewes (experiment 2) without any significant change in blood glucose concentrations. In comparison with the SO group the baseline-corrected areas under the insulin response curve were significantly (P < 0.05) smaller for the respective vagotomized groups for periods 1-2, 2-4 and 4-6 min (experiment 1) and 1-2 and 2-4 min (experiment 2) after presentation of food. Total area under the response curve for 10 min was significantly (P < 0.05) lower (experiment 1) and tended (P < 0.10) to be lower (experiment 2) for the vagotomized groups compared with that of the control groups. Direct electrical stimulation of the cervical vagus nerves raised plasma insulin concentrations to significantly (P < 0.05) higher levels in the SO ewes but not in the ADV ewes (experiment 3). It was also evident that in experiment 1, HADV did not have any additive effect over that achieved by ADV alone. These results indicate that the vagal innervation of the gut mediates insulin release during the cephalic phase of feeding in sheep. It is concluded that insulin secretion from the pancreatic -cells in response to either food-related reflex activation of the vagal nuclei in the hypothalamus or direct cervical vagus nerve stimulation is mediated through the vagal efferent fibres carried in the abomasal, pyloric and duodenal branches of the vagus nerves in sheep.