Although high school teacher evaluations of student non-cognitive attributes, such as leadership ability, are frequently considered in selecting applicants for competitive colleges, little empirical research exists on the reliability, validity, and bias of teacher ratings. The United States Military Academy at West Point requires that a set of school officials rate applicants on 10 non-cognitive items designed to measure potential for leadership in the highly structured Academy environment. Measures of leadership potential and student self-reports of athletic participation and extracurricular activities are also considered. Leadership performance at the Academy is assessed by cadet officers, Army officers, and upperclassmen. For a sample of 15,557 applicants to West Point, teacher ratings of leadership potential were related to measures of extracurricular participation and academic performance. Such ratings were found to be unbiased with respect to race and sex. Measures of leadership performance in the first two years at West Point correlated significantly with teacher ratings. Other alternative measures were not as valid as the teacher ratings. (NRB)