WHILE ATTEMPTS HAVE BEEN MADE TO CONSTRUCT MEASURES WHICH HAVE VALIDITY FOR PREDICTING ACADEMIC SUCCESS IN COLLEGE, MOST OF THESE TESTS HAVE BEEN DEVELOPED FOR NATIONWIDE TESTING PROGRAMS AND MAY NOT BE AS COMPLETELY VALID FOR MEASURING THE APTITUDE OF MINORITY GROUP HIGH SCHOOL STUDENTS WHO HAVE RESTRICTED ENVIRONMENTAL BACKGROUNDS. TO TEST THE PREDICTIVE VALIDITY OF ONE OF THESE TESTS, A STUDY COMPARED THE CORRELATIONS OF THE SCHOLASTIC APTITUDE TEST (SAT)-VERBAL AND SAT-MATHEMATICAL SCORES WITH THE FRESHMAN GRADE POINT AVERAGES OF NEGRO AND WHITE GROUPS. THE NEGRO POPULATION WAS SELECTED FROM THREE SOUTHERN NEGRO STATE COLLEGES, AND THE WHITE POPULATION WAS SELECTED FROM THREE NON-NEGRO COLLEGES WHO HAD THE LOWEST AVERAGE SCORES ON BOTH FORMS OF THE SAT. ONLY THREE ANALYSES WERE MADE. IT WAS FOUND THAT THERE WERE NO SIGNIFICANT DIFFERENCES IN PREDICTIVE VALIDITIES ASSOCIATED WITH RACE, SUBTEST, AND YEAR. HOWEVER THE INTERACTION OF SEX WITH RACE DID SHOW SOME DIFFERENCES. THE FRESHMAN GRADE POINT AVERAGE WAS FOUND TO BE MOST PREDICTABLE AMONG WHITE WOMEN, AND THE WHITE MEN WERE FOUND TO BE THE LEAST PREDICTABLE GROUP. THIS DOCUMENT WAS PRESENTED AT THE AERA MEETINGS IN 1967. (DK)