THE EPIDEMIOLOGY of tuberculosis may be investigated in four ways. The most frequent method is to determine mortality rates from information contained in death registrations. In this manner large numbers of cases may be obtained, but there are two sources of error in such data to which little attention has been given. Practically all causes of death are based on clinical and not on pathologic diagnosis, and pathologists are aware of the inaccuracies which are apt to be present in such data. Even if the clinical diagnosis of the cause of death is correct, this does not exclude the possibility that active pulmonary tuberculosis was present and that tubercle bacilli were being excreted—an important condition when the epidemiology of the disease is considered. A second procedure is the tuberculin survey, from which the rate of infection may be accurately determined and from which suggestions arise as to where to search