In two studies, we investigated how skilled readers use contextual and morphemic information in the process of incidental vocabulary acquisition during reading. In Experiment 1, we monitored skilled readers' eye movements while they silently read sentence pairs containing novel and known English compound words that were either semantically transparent (e.g., "milkshake," "drinkblend") or opaque (e.g., "cocktail," "deskdoor") in informative and neutral sentence contexts. In Experiment 2, we included a postreading vocabulary test following self-paced reading, which allowed for an examination of word learning success. We found that readers showed processing time advantages for novel transparent compounds in informative contexts, when contextual and morphemic information converged on a meaning. Conversely, these readers showed processing time disadvantages for novel opaque compounds in informative contexts when contextual and morphemic information conflicted as to the interpretation of the novel compound. Postreading vocabulary test performance revealed that readers retained new word meanings after a single exposure, and retention rates were significantly improved as a result of combining contextual and morphemic cues. Further, readers spent more time reading sentences containing novel words for which they subsequently correctly identified the meaning of the novel word. The data suggest that skilled readers automatically decompose novel compound words, that readers' ability to derive word meanings for novel English compounds is aided by pooling contextual and morphemic sources of information, and that readers' monitoring of their comprehension when reading texts containing novel words leads to further gains in vocabulary knowledge. (Contains 4 notes and 7 tables.)