The validity of prediction equations based on American College Testing (ACT) Program test scores and high school grades is investigated. Presented are separate results for prediction equations based on ACT scores only, on high school grades only, and on both kinds of prediction jointly. Also included is a discussion of the relationship between equations and the length of time they have been in use. The study is based on data collected from a random sample of 260 colleges participating in the ACT Research Services from 1972-73 to 1976-77. The accuracy of prediction equations based on the two variables was found to be quite stable over time. A similar stability was noted in the accuracy of grade predictions for males and females separately. Grade predictions based on ACT scores only showed slightly greater stability over time than those based on grades only. It is concluded that despite the changes some colleges experience in student abilities, curricula, and grading practices over time, freshman grade averages can be accurately predicted using equations as old as four years. ACT policy is to update equations at least every three years. Data tables and statistical analyses are briefly discussed, and a brief list of references is included. (Author/MSE)