Abstract In 131 patients on a medical service and 97 patients on a surgical service, in whom a diagnosis of hepatobiliary disease was verified in the hospital, the diagnostic value of routine liver tests performed soon after admission was evaluated by stepwise discriminant analysis. By measurements of alanine aminotransferase, alkaline phosphatases, gamma globulin, prothrombin time, bilirubin, and albumin, half of the medical patients were correctly classified into one of seven diagnostic categories. Aminotransferase contributed most to the classification, being twice as effective as random allocation. Decreasing the number of diagnostic categories to three (hepatitis, fatty liver, and chronic liver disease) increased the frequency of correct allocation to 80%. The allocation of all the patients to seven medical and four surgical diagnostic categories by means of four tests (aminotransferase, alkaline phosphatases, prothrombin time, and bilirubin) was significantly improved by each step, with a misclassification rate of 55% when all tests were used. A reduction of the diagnostic groups to five (hepatitis, fatty liver, chronic liver disease, duct obstruction and tumor) increased the frequency of correct allocation to 63%. The analysis demonstrates the limited diagnostic effectiveness of routine liver tests when used alone. The absolute discrimination values depend on the a priori frequencies of the diagnostic groups investigated, and therefore may vary from time to time and from place to place.