Purpose: The purpose of this study was to determine the acute dose-response relationship of classroom exercise breaks with executive function and math performance in 9- to 12-year-old children by comparing 5-min, 10-min, or 20-min classroom exercise breaks to 10 min of sedentary classroom activity. Method: This study used a within-subjects experimental design conducted in the spring of 2012. Ninety-six 4th- and 5th-grade students in 5 classrooms in South Carolina were randomized to receive each of 4 treatments: 5-min, 10-min, or 20-min exercise breaks or 10 min of a sedentary lesson led by research staff. Students completed the Trail-Making Test, an Operational Digit Recall test, and a math fluency test immediately before and after each condition. Planned linear contrasts were used to compare posttest scores between conditions using a repeated-measures mixed model, adjusted for gender, classroom, and the time-varying pretest scores. Potential effect modifiers were added as interaction terms. Results: Math scores were higher after the 10-min and 20-min exercise breaks compared with the sedentary condition (d = 0.24, p = 0.04, and d = 0.27, p = 0.02, respectively), and an interaction was observed with gender, IQ, aerobic fitness, and lower engagement in some of the conditions. There were no improvements in executive function tasks. Conclusions: A 10-min and 20-min classroom exercise break moderately improved math performance in students compared with a seated classroom lesson.