This study examines the predictability of GRE reading item difficulty (equated delta) for three major reading item types: main idea, inference, and explicit statement items. Each item type is analyzed separately, using 110 GRE reading passages and their associated 244 reading items; selective analyses of 285 SAT reading items are also presented. Stepwise regression analyses indicate that the percentage of GRE delta variance accounted for varied from 20% to 52% depending upon the item type. Details of item predictability were explored by evaluating several hypotheses. Results indicated that (1) multiple-choice reading items are sensitive to variables similar to those reported in the experimental literature on comprehension, (2) many of these variables provide independent predictive information in regression analyses, and (3) substantial agreement between GRE and SAT reading predictability was found.