The purpose of this study was to determine whether it is both possible and cost-effective to revise middle-difficulty Graduate Record Examinations (GRE) discrete items in order to produce items of higher or lower difficulty. The basic procedure was to select items of a given difficulty and, by revising the distractors, make them easier or more difficult. It was found that it is significantly easier to increase the difficulty of middle-difficulty items than to reduce the difficulty of such items and that the difficulties of antonyms and analogies are much easier to manipulate than those of sentence completions. The evidence also suggests that producing harder analogies and antonyms by revising items in this manner would be a cost-effective procedure. Nine appendixes present definitions of GRE content categories, examples of revised items, statistical analysis results for revised items and the tests, pretest yields, and cost figures. Three tables illustrate the discussion. (SLD)