This article reports literacy outcomes for a 2-year longitudinal student sample and a combined longitudinal and "in-mover" (i.e., those students who moved into the study schools between the initial pretest and the second-year posttest) sample, both of which were nested within 38 schools. Through the use of a cluster randomization design, schools were randomly assigned to implement Success for All or control methods. Hierarchical linear model analyses involving the longitudinal sample revealed statistically significant school-level effects of assignment to Success for All on three of the four literacy outcomes measured. Effects were as large as one quarter of a standard deviation--a learning advantage relative to controls exceeding half of a school year. Impacts for the combined longitudinal and in-mover sample were smaller in magnitude and more variable. The results correspond with the Success for All program theory, which targets school-level reform through multiyear sequencing of intensive literacy instruction. (Contains 5 tables and 3 notes.)