To determine the most accurate means of estimating the expected grade or reading level of 61 sixth grade students, three expectancy formulas were compared to a custom regression equation from a user friendly microcomputer program. The expectancy formulas, based in part on student I.Q. and chronological age, and the regression equations were compared with students' actual achievement level as measured by the California Achievement Test. Results indicated that the regression equation had twice the predictive power of the generalized expectancy formulas and could easily be developed by teachers for their own classes. (MM)