Two experiments investigated the effects of preceding sentence context on the naming times of sentence completion words in third-grade children and college students. In the first study subjects were shown incomplete sentences with four types of target words: best completions; semantically and syntactically appropriate, but less likely completions; related associates of the best completion words; and anomalous words. In the second experiment the target words consisted of those that were both semantically acceptable in the sentence and semantically related to the best completion words; words which were semantically acceptable in the sentence but not strongly related to the best completion words; and semantically unacceptable completion words that were not strongly related to the best completion word. After the incomplete sentence was flashed on a screen, the target word appeared. Students were told to read the target word aloud as rapidly as possible. Results showed that contextual facilitation effects were observed both for highly predictable and for semantically appropriate (but less likely) completions. In both cases, the facilitation effect was much greater for children than for adults. Children also exhibited contextual interference for anomalous word completions, whereas no such effects were observed in adults. (Materials used in the study are appended.) (HOD)