A randomized controlled trial was used to evaluate the effects of a supplemental mathematics intervention that emphasized fluency building for computations and procedures. All fourth- and fifth-grade English-speaking students from a single school district who were participating in general education mathematics instruction and had a 2009 year-end accountability score were included in the study (N = 537). Assignment to intervention or control conditions occurred at the classroom level with roughly half of classes being assigned to intervention and half being assigned to control conditions within each school. Outcome measures included the 2009 year-end statewide accountability measure in mathematics as well as three computation curriculum-based measurement probes of mathematics administered on three occasions during the school year. Implementation integrity data were collected in intervention classrooms via direct observation, teacher surveys, and monitoring of permanent products. Multilevel linear modeling (MLM) analysis was used to evaluate the effect of the intervention on student mathematics outcomes for all students and for students who were low performing at baseline. Intervention effects were detected at both grade levels (but not on all outcome measures). MLM was also used to evaluate the mediating effect of intervention implementation integrity on intervention effects. The integrity with which the intervention was implemented in intervention classrooms predicted end-of-year standardized measures of mathematics achievement and growth on curriculum-based measures. Implications for mathematics instruction in general and mathematics intervention within response to intervention models are discussed. (Contains 2 figures and 9 tables.)