Is the Armed Forces Qualifying Test (AFQT) a measure of achievement or ability? The answer to this question is critical for drawing inferences from studies in which it is employed. In this paper, we test for a relationship between schooling and AFQT performance in the NLSY 79 by comparing test-takers with birthdays near state cutoff dates for school entry. We instrument for schooling at the test date with academic cohort – the year in which an individual should have entered first grade – in a model that allows age at the test date to have a direct effect on AFQT performance. This identification strategy reveals large impacts of schooling on the AFQT performance of racial minorities, providing support for the hypothesis that the AFQT measures school achievement.