We attempted to strengthen an evidence-based, peer-mediated, first-grade reading program (First Grade Peer-Assisted Learning Strategies [PALS]) by modestly revising its content and adding a repeated-reading (RR) component. In a cluster-randomized trial, we conducted a component analysis of the revised program by creating two versions of it. "PALS+Fluency" represented the modified program with an RR component, whereas "PALS-Only" represented the same program without RR. We tested the efficacy of the two PALS versions together against controls and against each other to determine if peer-mediated RR had "value added." With moderator analyses, we further explored whether the PALS programs benefited weaker and stronger readers alike. Thirty-three first-grade classroom teachers (and 491 students) from eight urban schools were randomly assigned to the three study groups within their schools. The PALS-Only and PALS+Fluency programs ran for 22 weeks. Multilevel modeling showed that the combined effects of the two PALS programs on phonological awareness (PA), word reading, and reading fluency were superior to those of controls. Students' pretreatment PA moderated combined program effects on PA and word reading, indicating stronger effects for students with weaker PA. PALS-Only students improved PA skills (over PALS+Fluency students) with stronger effects for those with weaker pretreatment PA.