Assigning letter grades in a consistent manner to tests in large classes across semesters is problematic if absolute grading standards are used. It may be unreasonable to implement the usual standard-setting approaches recommended for large-scale criterion-referenced testing due to both time constraints and a desire to have criteria that appear uniform. However, percentage-correct grading standards cannot be fairly applied without adjustment to tests of differing difficulty. The suggestion is made that linear equating with an anchor test design may be an appropriate procedure for making the adjustment in many such circumstances. An example using real data from final examinations of an introductory social science course taken by 597 students in the winter and 609 students in the spring is examined. Apparently small differences in test difficulty are seen to yield large differences in the grades assigned when scores are put on a common scale. (Contains 2 tables and 10 references.) (Author/SLD)