The cephalic phase of pancreatic secretion in humans was investigated using modified sham feeding and a duodenal perfusion system. Studies performed in 5 normal volunteers were designed so that trypsin and bicarbonate outputs during sham feeding, with or without pretreatment with atropine, were compared to "maximal" pancreatic secretory response to exogenous stimulation with caerulein and secretin. The role of gastric acid entry to the duodenum in mediating cephalic responses was assessed by a comparison between outputs observed when gastric aspiration (congruent to 80% efficient) was used alone and when acid entry was completely abolished by combining gastric aspiration with cimetidine pretreatment. To evaluate the role, if any, of gut hormone release in the pancreatic secretory response to sham feeding, plasma gastrin and cholecystokinin concentrations were monitored throughout. Trypsin outputs during sham feeding were 31.9 +/- 10.45 kallikrein inactivator units per 30 min, equivalent to four times basal output and 92% of maximal, but were only 54% maximal in subjects pretreated with cimetidine. Atropine suppressed basal trypsin output and abolished the response to sham feeding (4.98 +/- 3.89 kallikrein inactivator units per 30 min). A modest increase in bicarbonate secretion during sham feeding (3.30 +/- 1.97 mmol/30 min versus basal of 0.68 +/- 0.74 mmol/30 min, p = 0.5) was not influenced by atropine but was abolished by cimetidine pretreatment. No significant changes in plasma gastrins were observed in these studies and plasma cholecystokinins remained undetectable throughout. We conclude that there is tonic vagal stimulation of trypsin secretion, and that sham feeding markedly increases trypsin output, which is augmented further by acid entry into the duodenum. There is no direct effect of cephalic stimulation on bicarbonate secretion or on gastrin or cholecystokinin release.