In this paper I propose and evaluate a grading standard design that allows for multiple grading standards across the grade distribution. I show that making it harder for a student to receive a higher grade when at a higher point in the grade distribution than to receive a higher grade when at some point lower leads to larger test score gains overall. Effects differ by race, gender, and at different points in the achievement distribution. Much of this analysis follows the method used by Betts and Grogger (2003) and I present my replications of their results for comparison purposes.