This study sought to determine which, if any, of the possible predictor variables based on relationships with college performance available at freshman entry would prove useful in selecting students for admission. A study of the relationships between four college performance variables and eight admission variables was conducted using seven freshman cohorts (summer and fall) at the University of South Florida (USF). Meaningful relationships with all four outcomes occurred for both high school grade point average (GPA) and class rank. Although simple relationships with outcome variables occurred for other variables such as sex, race/ethnicity, and test scores, most of their predictive capacity was already included in the GPA-based measures, and they added little to predictions. Even the very strongest relationships between predictors and outcomes failed to identify meaningful performance differences between any two adjacent scale points. Using a standardized 15-point scale (each point included roughly 7% of all students) across GPA. Class rank, ACT Assessment, and Scholastic Assessment Test scores, attrition differences between adjacent scale points at the maximum were about 2.4% and those for meaningful outcomes (graduation and retention to the second year) were generally between 0.5% and 2.0%. Even without considering the issue of measurement error, these small differences in outcomes from scale point to scale point show the fantasy of selecting specific admission cutoffs. When one sets a cutoff, no difference in outcomes will occur for the students immediately above and below the cutoff in at least 97.6% of cases, and almost always in 89 to 99% of cases. T. Mortenson (1999) indicates that affluence related far more with graduation and retention and could reduce the incorrect decision rate from 98% to perhaps 70% of rejected applicants. (Author/SLD)