When do item adaptations veer from their intent and, instead of increasing access, modify the construct being measured? This study analyzed early elementary student achievement data from a statewide field test containing both standard and adapted science items. Four student groups were included in this analysis: English language learners, students with learning disabilities, students who are deaf or hard of hearing, and a control group of non-individualized education program native English speakers. This study included a qualitative evaluation of the items by experts with backgrounds in working with students with disabilities. Small sample sizes and restricted numbers of items available made generalizations of the results difficult, but findings suggest unexpected focal group differences in performance on the adapted items. The authors suggest that attention needs to be paid to whom items are adapted for, and steps need to be taken during item development to ensure that changes have the intended effect. (Contains 6 tables and 7 figures.)