The stud}' examines validities of intelligence test scores for predicting achievement for 7,000 males over a 2i-year period of military service in Israel. Specifically, the study examines whether differential test validity occurs in a very heterogeneous population by comparing validities for three ethnic subgroups, namely, those born in Israel, in Western, or in Eastern countries. Considerable differential validity was found. However, within separate educational levels, this effect was largely eliminated. It is suggested that differential ethnic validity, when it occurs, may sometimes be usefully circumvented by using education as a predictor. The theoretical and practical implications are discussed.The question of test fairness-whether test scores represent potential to the same extent for different ethnic groups-has received considerable attention in recent years. On the face of it, the content of tests commonly used for personnel selection would not seem to be completely equivalent in their relevance for different groups (Vernon, 1969); particularly in the social-economic and ethnic circumstances of the United States, test fairness has become more than a theoretical problem. Clearly, many influences are involved, and "the extent of such influences is a matter for empirical determination rather than for a priori assumption" (Anastasi, 19SO, p, 77). Consequently, manifestations of nonequivalence have been extensively sought in recent years and have merited special attention in two Annual Review of Psychology articles on personnel selection (Ash & Kroeker, 197S, pp. 486-489; Bray & Moses, 1972, pp. S47-SSS).Because of the breadth of the problem,The material for this research was accumulated while the author was employed by the Research Department of the Israel Defence Forces Manpower Branch, to whom thanks are expressed for permission to publish it. The findings and opinions expressed are those of the author and not necessarily those of the Israel Defence Forces.Thanks are also expressed to the Bar-Han University Computer Center, and particularly to M. Snyder, who developed the program for the comparison of regression lines and helped in interpreting its results.Requests for reprints should be sent to M.