Mathematics Recovery (MR) is designed to identify first graders who are struggling in mathematics and provide them with intensive one-to-one tutoring. We report findings from a 2-year evaluation of MR conducted in 20 elementary schools across five districts in two states. The design allowed for the estimation of the counterfactual growth trajectory based on those students randomly assigned either to a tutoring cohort with a delayed start or to a wait list. Results demonstrate strong end of first grade effects on a diagnostic measure developed by MR and weak to moderate effects (effect size, 0.15-0.30) on measures administered by external evaluators. By the end of second grade, no significant effects were found on any measures. Practical and research implications are discussed. (Contains 7 tables, 3 figures, and 5 notes.)