This study provided an exploratory investigation of item features that might contribute to a lack of invariance of item parameters for the Test of English as a Foreign Language (TOEFL). Data came from seven forms of the TOEFL administered in 1989. Subjective and quantitative measures developed for the study provided consistent information related to the model-data fit of TOEFL test items. For TOEFL Sections 1 and 2, items that were pretested before 1986 exhibited poorer model-data fit than items that were pretested after 1986. For Section 3, reading comprehension, model-data fit appeared to be related to changes in the relative position of items within the sections from the pretest to the final form administrations. Based on the results of the study, it was recommended that the TOEFL program investigate the feasibility of not using pretest item response theory statistics for items pretested before 1986 for Sections 1 and 2 and that guidelines be developed for test developers to use with reading comprehension items to limit change in relative positions of items in the test from pretest to final form administrations. Two appendixes give rules for judging item-ability regression plots and sample item ability regression plots. (Contains 4 figures, 11 tables, and 18 references.) (Author/SLD)