How does the representation of boy and girl exemplars in curricular materials affect students' learning? We tested two competing hypotheses about the impact of gender exemplar on learning: First, in line with Social Learning Theory, children might exhibit a same-gender bias such that they prefer to learn from exemplars that match their gender (H1). Second, consistent with research on children's stereotypes about gender and math (e.g., associating boys with math competence), children might prefer to learn from exemplars who match their stereotypes about who is good at math (H2). We tested these hypotheses with children in middle school (N = 166), a time of development in which stereotypes are well-engrained, but before gender differences in math achievement appear. Children viewed two distinct math strategies, each presented by a boy or girl exemplar. We then examined which strategy children employed on a subsequent math test as well as their perceived similarity to the exemplars and their awareness or endorsement of gender-math stereotypes. Children did not preferentially learn from same-gender exemplars. However, children with stereotypes associating boys with math were more likely to learn the more difficult strategy when it was presented by a boy exemplar than children who did not associate boys with math. The results of this study provide valuable insight into how children's stereotypes impact their real-world learning.