Theoretical models of word recognition suggest that knowing what a word means makes it easier to learn how to decode it. We tested this hypothesis with at-risk young students, a group that often responds poorly to conventional decoding instruction in which word meaning is not addressed systematically. A total of 53 first graders received explicit instruction on how to decode 32 words. Researchers also taught them the meanings of a random selection of these words. Overall, decoding instruction seems to have been effective: Students read words more accurately and more quickly at posttest than at pretest. They were also able to learn the meanings of many words when this aspect was addressed in instruction. However, accuracy or fluency of word decoding did not vary according to whether word meaning was known. We examine theoretical and practical implications of these findings.