Study was made of the efficiency with which mechanical classification and correlation of data might assist in the differential diagnosis of hematological diseases. Data of 80 patients with hematological conditions were correlated with data characteristic of 26 hematological diseases. Data of each case were correlated with the data of the diseases simultaneously. The procedure made it possible to tell whether a hospital case contained data identical with those characteristic of any disease. The information returned also told whether enough findings in the hospital case were present to make a diagnosis of that dis case. If not enough data were present to make a diagnosis, the information returned indicated which further tests were needed. Although it was possible to end the correlation procedure with no diagnosis, it was not possible to end with the wrong diagnosis, if the data of the hospital case were correct. It is believed that the efficiency with which correct diagnoses were identified in this study justifies further evaluation of these methods in correlating data of this type.