The main purposes of this study were to test the effects of teaching at-risk 4th graders to provide explanations for their mathematics work and examine whether those effects occur by compensating for limitations in cognitive processes. We randomly assigned 212 children to 3 conditions: a control group and 2 variants of a multi-component fraction intervention. Both intervention conditions included 36 sessions, each lasting 35 min. All but 7 min of each session were identical. In the 7-min component, students were taught to provide high quality explanations when comparing fraction magnitudes or to solve fraction word problems. Children were pretested on cognitive variables and pre/posttested on fraction knowledge. On accuracy of magnitude comparisons and quality of explanations, children who received the explaining intervention outperformed those in the word-problem condition. On word problems, children who received the word-problem intervention outperformed those in the explaining condition. Moderator analyses indicated that the explaining intervention was more effective for students with weaker working memory, while the word-problem intervention was more effective for students with stronger reasoning ability.