The amino acid consumption test (AACT) has been proposed as a simple tubeless test of pancreatic function, but studies of its diagnostic accuracy have produced conflicting results. Eighty-three consecutive patients with clinical suspicion of pancreatic disease underwent pancreatic stimulation for 1 h with an intravenous infusion of cerulein (50 ng/kg/h); the total plasma amino acid concentration was measured at -15, 0, 30, 45, and 60 min of infusion. The maximal percentage decrease in plasma amino acid concentration during cerulein infusion was taken as an index of pancreatic function. In addition, patients had pancreatic function assessed with the pancreolauryl test (PLT). Of the 83 patients studied, 24 were found to have chronic pancreatitis and four pancreatic cancer; the remaining 55 had various nonpancreatic digestive disorders. Pancreatic function, as assessed by the PLT, was impaired in 22 of the 28 patients with pancreatic diseases, and it was normal in all but four patients with nonpancreatic disorders. Cerulein infusion caused a decrease in total plasma amino acid concentration that was generally more pronounced in patients with nonpancreatic diseases (maximal percentage decrease: median, 14%; range, 4-28%) than in those with pancreatic diseases (maximal percentage decrease: median, 9%; range, 0-21%) (p < 0.001). Using a cutoff of a 14% amino acid decrease, the sensitivity of the AACT was 89% and the specificity 53% (diagnostic accuracy, 65%); with a cutoff of 12%, the sensitivity was 75% and the specificity 69% (diagnostic accuracy, 71%). The sensitivity of the PLT was 79% and the specificity 93% (diagnostic accuracy, 88%). The results indicate that the sensitivity of the AACT is relatively high, but they show that the specificity is low, making the test unsuitable for clinical use, at least in its present form.