OBJECTIVE To study the effect of meal composition on pancreatic polypeptide release during modified sham feeding. METHODS In random order and on separate occasions, isocaloric, isothermic, isoosmotic, homogenized meals (1050 kJ; 250 kcal) either rich in fat (walnuts; 64 g fat, 7 g protein, 15 g starch per 100 g), protein (codfish, 1 g fat, 23 g protein per 100 g) or carbohydrates (bananas; 22 g starch, 1 g protein per 100 g) were sham-fed for 30 min by tasting and spitting out the meal. The plasma pancreatic polypeptide response was monitored by radioimmunoassay at 10 min intervals from 20 min before to 120 min after modified sham feeding. METHODS Department of Gastroenterology and Hepatology of a University Hospital. METHODS Seven healthy volunteers: 3 male, 4 female; age 45 (range 30-77) years. RESULTS Integrated plasma pancreatic polypeptide responses to modified sham feeding of codfish (1088 +/- 395 pM*120 min; P < 0.05) and walnuts (1200 +/- 542 pM*120 min) were distinctly higher (P < 0.05) than to modified sham feeding of bananas (-390 +/- 291 pM*120 min). CONCLUSIONS These results demonstrate that the pancreatic polypeptide response to modified sham feeding is dependent on the composition of the meal.