Prediction of the difficulty (equated delta) of a large sample (n=213) of reading comprehension items from the Test of English as a Foreign Language (TOEFL) was studied using main idea, inference, and supporting statement items. A related purpose was to examine whether text and text-related variables play a significant role in predicting item difficulty. It was hypothesized that multiple-choice reading comprehension tests are sensitive to many sentential and discourse variables and that many of these variables contribute significant independent variance in predicting item difficulty. The majority of the sentential and discourse variables identified in the literature were found to be significantly related to item difficulty within TOEFL's multiple choice format. A significant relationship was found between item difficulty and text and text-related variables, supporting the claim that multiple-choice items yield construct valid measures of comprehension. For the full sample of 213 items, with the equated delta the dependent variable, 33% of the item difficulty variance could be accounted for by 8 variables. Five tables present analysis details, and an appendix gives correlations and regressions for each of three item types. (Contains 43 references.) (SLD)