When test questions are not intended to measure language skills, it is important to know if language is an extraneous characteristic that affects item performance. This study investigates whether certain stylistic changes in the way items are presented affect item performance on examinations for a health profession. The subjects were medical residents taking a multiple-choice in-training examination. Twenty items were selected from the pretest section of the examination to cover three stylistic manipulations: (1) abbreviations; (2) symbols; and (3) generic and proprietary drug names. Every item was tested in two versions (unedited, edited) and these item pairs involved the same stylistic change for each pair. A Bonferroni test of significance was applied to the differences across forms for each item pair. A two-way multivariate analysis of variance yielded essentially the same results as the two-way analysis of variance. No significant differences in performance were noted for individual item pairs or for type of stylistic manipulation. However, a test form effect was noted on one of the two replications. Test developers need to be aware that even if there is no difference in statistics on an item level, differences might exist on a test form level. (Author/PN)