The literature is replete with references to a disproportionate reliance on admission test scores (e.g., the ACT or SAT) in the college admissions process. School-reported reliance on test scores and grades has been used to study this question, generally indicating relatively equal reliance on the two, with a slightly higher endorsement of grades. As an alternative, we develop an empirical index of relative reliance on tests and grades, and compare school-reported estimates with empirical evidence of relative reliance. Using a dataset from 174 U.S. colleges and universities, we examine the degree to which applicants and enrolled students differ on the SAT and on high school GPA in each school, and develop an index of empirical relative reliance on test scores vs. grades. We find that schools tend to select on test scores and high school grades relatively equally, with the empirical reliance index showing slightly more reliance on test scores and school-reported reliance estimates showing slightly more reliance on grades.