Heretofore the roentgenologic diagnosis of pituitary tumors has been limited to those growths far enough advanced to cause the typical appearance which has come to be associated with sellar changes in this type of lesion. The diagnosis of early lesions has not been made, owing largely to the difficulty in distinguishing between the normal sella and the sella with borderline enlargement. This problem has been complicated by the fact that on routine measurement 10 to 20 per cent of the acidophilic tumors do not show enlargement, and that the basophilic tumors thus far studied, with rare exceptions, have been microscopic in size. Changes in these latter cases are limited to the alterations of calcium architecture in the cranial vault. The chromophobe tumors, occurring more frequently than the others, have produced changes in all cases thus far diagnosed. In a number of cases of enlarged sella there has been no clinical evidence of the disease. The problem has been further complicated by many borderline enlargements found on routine examination. It became apparent that we needed to make a careful review of our methods of measurement to determine whether an accurate study of the composites of the pituitary fossa and adjacent bony structures could be made on routine skull films. Analysis of our routine procedure revealed (1) that it did not furnish a satisfactory method of measuring the size of the pituitary fossa, and (2) that measurements were limited to the size as seen in the lateral view. In working out a satisfactory method of measurement of the sella we have been stimulated by Dr. L. M. Hurxthal, who has been particularly interested in determining the size of the pituitary fossa in the normal individual according to age, weight, and height, so that abnormalities in size might be demonstrated in certain glandular disturbances as well as in cases of pituitary tumor. A more satisfactory technical method has also been devised whereby we are able to study the dorsal and anterior elinoids, the dorsum sellae, and the floor of the sella without detracting too much from the demonstration of other portions of the cranial vault. This has been done without increasing the number of films used in study of the skull. The roentgenograms are obtained with the patient in the upright position. A set of lateral and postero-anterior stereoscopic films is taken for routine study in the following manner. For lateral stereoscopic views of the skull, in order not to show distortion of the floor of the sella, one film must be taken in the exact lateral position. To do this, the jaws must be superimposed and the central beam must be made to pass through the floor of the sella by directing the beam 2 cm. posterior to the outer canthus of the eye just above the zygomatic arch. The second film is taken with a cephalad shift of 3 1/2 inches with a 5-degree caudal tilt.