To increase specificity in the clinical determination of amylase, the differential activities of human pancreatic and salivary amylases toward two different substrates—a Lintner soluble starch (LSS) and an insoluble, dyed starch, Amylose Azure (AA)—were investigated. A ratio of initial-velocity activities of Lintner soluble starch to Amylose Azure (LSS/AA) for salivary amylase was at least twice as great as the LSS/AA ratio obtained for pancreatic amylase. Statistical analysis of paired saliva samples and secretin-pancreozymin stimulated pancreatic juices revealed a significant (p < 0.01) difference between the mean LSS/AA activity ratios of these two biologic fluids. Even greater differences were found between the ratios for crystalline salivary amylase compared to the amylase contained in purified preparations from human duodenal fluid and pancreatic homogenates. A method to differentiate elevated serum amylase resulting from parotitis or pancreatitis is demonstrated.