The current study uses a crossed random-effects item response model to simultaneously examine both reader and word characteristics and interactions between them that predict the reading of 39 morphologically complex words for 221 middle school students. Results suggest that a reader's ability to read a root word (e.g., "isolate") predicts that reader's ability to read a related derived word (e.g., "isolation"). After controlling for root-word reading, results also suggest that the remaining variability in derived-word reading can be explained by word and reader characteristics. The significant word characteristics include derived-word frequency and root-word frequency but not morpheme neighborhood size, average family frequency, number of morphemes, or semantic opaqueness. The significant reader characteristics include morphological awareness and vocabulary knowledge but not reading comprehension. Only phonological and orthographic-phonological opaqueness interacted with the effect of root-word reading, suggesting that students were less able to apply root-word knowledge when the root word changed phonologically (with or without an orthographic change) in the larger derived word. Discussion is included regarding how findings from this study inform the development of models of word reading for adolescents. (Contains 3 tables, 3 figures, and 2 notes.)