The purpose of this study was to examine word learning efficiency in at-risk first grade students (N = 93) participating in a yearlong study evaluating a multicomponent intervention targeting word reading and decoding skills. As part of each intervention lesson, students participated in a 1 to 3-min sight word reading activity in which high-frequency words were read from a list until mastered, at which point the word dropped off the list. This study explored factors predicting the number of exposures required for item reading mastery (N = 145 words). Specifically, we explored how the number of word exposures required to reach mastery varied as a function of linguistic features of the words and cognitive characteristics of the students. Using item-level crossed-random effects models, we found students required an average of 5.65 exposures for mastery, with word features representing word length, vocabulary grade, and imageability being significant predictors of learning efficiency. We also found a significant interaction between pretest word reading skill and imageability of a word, with this semantic feature being especially important for the poorest readers. Results indicate that in the absence of typical word recognition skills, poor readers tend to rely on other sources of information to learn words, which tend to be related to the semantic features of words.