In previous studies we found that in healthy subjects, 5 and 10 g of a partially purified amylase inhibitor delayed and decreased starch digestion and reduced postprandial plasma glucose after a starch meal but produced diarrhea in two of six and four of six subjects, respectively. Thus, we wondered whether lower doses of the inhibitor, when given with a meal that contained protein and fat as well as carbohydrate, would have the same effect on carbohydrate tolerance without causing diarrhea. Eight healthy subjects were randomized to receive 2.0 or 2.9 g of the inhibitor with a 650-calorie meal that contained carbohydrate, fat, and protein. In comparison with a placebo, ingestion of 2.9 g, but not 2.0 g, of the inhibitor significantly reduced postprandial increases in plasma glucose (P less than 0.05), C peptide (P less than 0.03), and gastric inhibitory polypeptide (P less than 0.008). Similarly, 2.9 g of the inhibitor in comparison with 2.0 g was associated with more carbohydrate malabsorption and more breath hydrogen excretion. Because the carbohydrate malabsorption observed with the 2.9-g dose was similar to that with the previously tested 5- and 10-g doses of the inhibitor but diarrhea was less frequent, impurities in the partially purified preparation may, in part, have been responsible for these adverse effects. We conclude that 2.9 g of the amylase inhibitor given with a meal that contains a mixture of nutrients is effective in increasing carbohydrate tolerance without causing diarrhea. Therefore, this dose is appropriate for use in studies to determine whether the inhibitor has a beneficial effect in patients with diabetes mellitus or obesity.