This study compared a performance-based, culture-free, learning test with a psychometric test battery, the Armed Services Vocational Aptitude Battery (ASVAB), for assessing individual differences among high-school students. Ethnic background (white-nonwhite) and sex were additional parameters. While negligible differences were found on the learning test between groups differentiated on ethnic background or sex, significant differences favoring white and males were found on the Armed Forces Qualification Test (AFQT)--a general aptitude composite formed from the ASVAB and used to derive cutting scores for entry into the Armed Services. The best predictor of the learning test score among the 12 tests of the ASVAB and the AFQT composite was the General Information (GI) test. The AFQT was also a reliable predictor, but only for whites. Analysis of the relationships suggested that the learning test was measuring a portion of the large, general intelligence component known to underlie the ASVAB tests and composites. The learning test appeared to measure this pervasive factor better than the AFQT over the entire range of abilities found in this sample. The results suggest that there may be many capable individuals being eliminated in selection programs using psychometric cutting scores of general cognitive ability or intelligence. Those being adversely impacted in this age group are not only the culturally disadvantaged and persons whose first language is not English, but females in general. (Author)