Playing board games and other math activities can provide young children with opportunities to develop their math skills. However, it is critical to understand for whom these activities may be most beneficial. In two studies, we examine the extent to which foundational cognitive skills moderate the effects of playing math games on math skills. In Study 1, we look cross-sectionally at the association between parents' frequency of math activities with their 3- to 4-year-old children (N = 124) and children's math achievement, examining the extent to which children's skills moderate this relation. We find that frequent math activities are only associated with better math performance for children with better number knowledge. In Study 2, we test this experimentally by randomly assigning parents and children (N = 76) to play with a number-related board game, an active control board game, or a business-as-usual control group. Controlling for number knowledge, inhibitory control, and vocabulary at pretest, no differences in math skills at posttest were observed between the training groups. However, a significant interaction emerged between training group assignment and number knowledge, such that children with higher pretest number knowledge had higher posttest math scores when assigned to the number board game condition compared to the two control conditions, but no differences among conditions were seen for children with lower number knowledge. Collectively, these findings suggest that math activities may be most beneficial for math skills when children have stronger number knowledge and underscore the need for tailoring activities to children's current skill level.