PREDICTIVE diagnosis of development is possible because behavior matures in an orderly fashion, but it is not a simple matter. Unlike most other diagnostic problems in medicine, disturbances in neuropsychologic functioning involve integration on many different levels, from the biologic to the sociocultural. Analysis of something as complex as behavior thus is not without difficulties, and extensive training in making, and interpreting, observations is necessary for accurate diagnosis. The number of children in whom abnormal development is suspected is enormous, the need for service great and the supply of specialists limited. Under these circumstances an adequate screening procedure which will detect infants in need of referral for a complete diagnostic evaluation is of great value. A number of devices and lists of behavior have been offered; when they have been tested at all it has not been for their correlation with an examination that has predictive ability. They have consisted for the most part of a few behavior patterns at widely separated ages and most have fallen into disuse because of their apparently unsatisfactory nature. CRITERIA FOR AN ADEQUATE SCREENING DEVICE In evaluating development one is concerned with both present and future functioning, and the complete examination from which the screening inventory is derived must be an adequate predictor of later abnormality. First, then, it must have validity. Since this predictive value can be determined only by longitudinal studies, such data must derive from follow-up correlations between complete infant examinations and those at later ages. Further, for prediction as well as for present management, intellectual defect and motor disability must be distinguished.