The purpose of this study was to (a) examine the underlying assessment structure of the Derivational Morphology Task (DMORPH) and (b) investigate the relation of the DMORPH to vocabulary and reading comprehension outcomes with a sample of struggling adult readers. Specifically, participants included 218 struggling adult readers enrolled in adult literacy classes. We used item-level analyses to evaluate the underlying structure of the DMORPH. Items with phonological (e.g., "music" to "musician") and non-phonological transformations ("teach" to "teacher") were examined in relation to adult literacy students' vocabulary and reading comprehension skills. A bifactor model was the best fit to the data, suggesting that the DMORPH measured a single factor of derivational morphological awareness with some variation due to phonological and non-phonological change items. Follow-up analyses revealed that the DMORPH can essentially be considered unidimensional, which justified the use of a single scoring system for the DMORPH with adult literacy students. However, after controlling for word reading and phonological awareness, the phonological change items uniquely predicted vocabulary knowledge and reading comprehension, whereas the non-phonological change items were not significant. The results support the structural validity of the DMORPH and the need to use both phonological and non-phonological change items with adult literacy students. The present findings also provide insight into potential intervention targets for instructors in adult literacy programs who are interested in improving students' vocabulary and reading comprehension skills.