Patients with clinically diagnosed hearing losses were given the verbal tests and threshold measurements used in the USPHS Health and Nutrition Examination Survey for 1971–75. A computer program for categorizing hearing losses from recognition data [J. Acoust. Soc. Am.61, S5(A) (1977)] was applied to the results of the tests. permitting comparison between results of the ordinary clinical tests and computer‐aided diagnoses. The program estimates mixtures of losses whereas clinical tests other than hearing by bone conduction primarily diagnose the major component. For minor hearing losses, the speech test does not resolve unequivocal differences between conductive and sensorineural losses, but for higher threshold elevations it appears to resolve conductive, sensorineural, and recruiting components.