It has been claimed that there is inadequate statistical correlation of pure pelvimetry with obstetrical outcome (7). Moreover, the incidence of cesarean section in a group of patients with more favorable pelvic measurements has been found to be higher than in a group in which these measurements were less favorable (6). This is in striking contrast to the results obtained with the cephalopeloimetric methods of Chassar Moir (8), Caldwell, Moloy, and Steer (9), and Ball (1, 2), which indicate a definite correlation between the measured cephalopelvic difference and obstetrical outcome. All three of these methods are deficient in that their accuracy is lowest where it is most needed—namely at the mid-pelvis—when disproportion exists. Moreover, the borderline group, in which no roentgen prediction is possible on the basis of these methods, is too large. Indeed, it exceeds the group in which a diagnosis of disproportion can be made with assurance. A preliminary study of the Ball method (12) was sufficiently promising to warrant further investigation because, of the three methods, it showed the smallest borderline group. The object of the study to be reported here was to correlate the findings by this method with the obstetrical results by charting both on graphs, to analyze the deficiencies of the method, and to suggest ways and means of correcting them. Method of Investigation Three hundred and fifty consecutive cephalopelvimetric examinations performed on ward patients of the Sloane Hospital for Women were studied. All patients were radiographed in the upright position at or near term, as described by Ball and Golden (2). For correction and computation of measurements a special slide rule (11) was substituted for the nomogram (5) employed by these authors. The films, including an additional stereoscopic inlet view, were sent to the Department of Obstetrics prior to measurement by the radiologist. All measurements were made by one of the authors (G. S. S.) after delivery, when the films were returned to the X-ray Department. Thus the radiologist's interpretation in no way influenced the management of delivery, a feature which makes this study unique. The obstetricians, however, had the opportunity of measuring the films by any method of their choosing (usually not the Ball method) and were at liberty to extract any information from the films before delivery and without the help of the radiologist. Details for all patients with borderline and definitely unfavorable measurements and for all undergoing cesarean section or midforceps delivery were entered on graph sheets. To these were added a liberal cross section of patients who had been referred for cephalopelvimetry and found to have normal measurements and normal deliveries. A large number of normal cases were left uncharted in order to avoid crowding the graph sheets.