Plasma concentrations of human pancreatic polypeptide (HPP) parallel exocrine pancreatic secretion in response to stimulation with cholecystokinin. We determined prospectively the relationships among fasting HPP level, integrated HPP response to infusion of cholecystokinin, and output of trypsin and also the sensitivity, specificity, and predictive values of the fasting HPP level in the diagnosis of exocrine pancreatic disease. Our study group consisted of 19 patients with acute pancreatitis, 17 with chronic pancreatitis, and 25 with ductal adenocarcinoma of the pancreas and 27 control subjects. In the control patients and those with chronic pancreatitis, significant correlations were detected between HPP level and output of trypsin (P less than 0.001) in response to infusion of cholecystokinin and between fasting HPP and integrated HPP levels (P less than 0.004); no correlation was detected between HPP level and steatorrhea. The sensitivity, specificity, and negative and positive predictive values of the fasting HPP level for detection of either chronic pancreatitis or pancreatic cancer were similar and approximated 0.88, 0.67, 0.88, and 0.66, respectively. The HPP concentration had no value in detecting acute pancreatitis. Because the fasting HPP level has a high degree of negative predictability and is simpler to measure than the integrated HPP level or the output of trypsin, it may be a useful test in patients suspected of having either chronic pancreatitis or pancreatic cancer. A fasting HPP level of 125 pg/ml or greater could be used to exclude chronic pancreatitis or pancreatic cancer, but the finding of a value of less than 125 pg/ml necessitates use of other diagnostic tests for reliable determination of the presence of these diseases.