The ability of products of digestion to stimulate pancreozymin secretion in man was investigated using a bioassay procedure, based on duodenal perfusion, which quantified the total outputs of pancreatic enzymes evoked by intraduodenal stimuli under steady-state conditions. Patterns of response resulting from physiologic intraduodenal concentrations of test material were basal output (with isotonic saline), washout of enzymes (with dextrose, micellar fatty acid, and amino acids), and sustained output of enzymes (with amino acids and micellar fatty acid). The sustained secretion of pancreatic enzymes found during the 2nd hr of perfusion and subsequently was characteristic of pancreozymin-induced secretion. The enzyme output in response to a mixture of essential and nonessential amino acids was significantly higher than that evoked by micellar fatty acid and was comparable with that resulting from the maximally tolerated dose of pancreozymin given by vein. Perfusion with essential amino acids caused enzyme outputs comparable to those induced by perfusion with the original amino acid mixture, whereas perfusion with nonessential amino acids had no effect. When the essential amino acids were tested individually, only phenylalanine, methionine, and valine caused significant increases in pancreatic enzyme output; the effect of tryptophan was indeterminate. However, the pancreatic enzyme output was less in response to these three essential amino acids than to mixtures containing all of them.