To justify its labor force quality requirements, the Army sought to demonstrate an empirical link between Armed Forces Qualification Test (AFQT) scores and soldier performance. The first data set contained written and hands-on tests on several weapons systems from the Army's training schools. The second data set used the Skill Qualifications Tests administered by the Army. Performance measures were formed into a multivariate regression model, using an instrumental variables technique to correct for measurement error in the AFQT variable. Other explanatory variables were sex, race, education, Army experience, and training. The results demonstrated that a statistically significant and positive relationship existed between AFQT scores and performance measures. AFQT, a measure of trainability, was a significant predictor of performance in the Army. These findings were consistent across a wide range of Military Occupational Specialties (MOS). No other variables were consistently significant across all MOS in both data sets. (YLB)