The existence of differential prediction of two-year grade point average is reported for gender groups within programs of study at the University of Iowa. Academic records of all freshmen entering the University in 1978 in the fields of Business, Liberal Arts, Pre-Medicine, and those undecided as to major were analyzed with respect to American College Testing (ACT) Program Composite score and the ACT Mathematics Usage, English Usage, Social Studies Reading, and Natural Science Reading subtest scores, using a two-year cumulative grade point average as the criterion of academic success. Johnson-Neyman analyses indicate regions on the predictor score scales where differential prediction has practical impact within each program. A large portion of students is generally affected. A methodology is outlined for finding a prediction algorithm to reduce differential prediction while maintaining prediction potency. (Author/BW)