Treatment dose is an understudied aspect of treatment effectiveness. This study compared the frequency with which a small-group mathematics intervention was delivered weekly (i.e., four times, twice, once) with a control condition while controlling for total duration. 101 at-risk students in grades 2-4 were randomly assigned to a condition following universal screening and skill-based assessments. Multilevel modeling was used to evaluate final score and growth on three measures. Results suggested that for the most proximal computation measure, treatment sessions occurring four times weekly produced clear benefits. On the application measure, students in all treatment groups outperformed students in the control condition. For the most complex computation measure, frequency was not a useful predictor. Grade was a moderating variable.