This pilot study examined the efficacy of a Tier 2 first-grade mathematics intervention program targeting whole-number understanding for students at risk in mathematics. The study used a randomized block design. Students (N = 89) were randomly assigned to treatment (Fusion) or control (standard district practice) conditions. Measures of mathematics achievement were collected at pretest and posttest. Treatment and control students did not differ on mathematics assessments at pretest. A series of random-effects models were estimated to compare gains between treatment and control conditions. Gain scores of intervention students were significantly greater than those of control peers on a proximal measure of mathematics achievement. The role of a strong theory-of-change model in the development and evaluation of mathematics interventions is articulated. Implications for researchers and educators designing and delivering instruction for at-risk students in a response-to-intervention model are discussed.