This study examined whether the observed effectiveness of recasts is influenced by the type of outcome measure used and whether different aspects of working memory are differentially associated with learners' performance on the various outcome measures. The participants were 90 learners of English as a foreign language, who were randomly assigned to a recast, a nonrecast, and a control group. A pretest-posttest-delayed posttest design was employed to detect any improvement in the learners' knowledge of one usage of the English past progressive construction. Many-facet Rasch measurement and correlational analyses yielded two main findings. First, recasts generated the greatest gains on an oral production test, lesser gains on a written production test, and the least gains on a written grammaticality judgment test. Second, in the recast group, participants with higher reading spans achieved more development on the written tests, while those with higher digit and nonword spans showed greater improvement on the oral test. For the nonrecast group, no association was found between the working memory and developmental measures.