Unlike boys, when girls perform a math test, they may experience stereotype threat (ST), namely the threat of confirming the negative stereotype of women’s lower abilities in math. ST causes a decrement in girls’ math performance (Huguet & Régner, 2007). One means to counteract ST effects is to provide girls with role models. Marx and Roman (2002) showed that girls performed significantly better at a math test when it was administered by a competent woman rather than a competent man. A girl who is top of the class in math can also be a role model for her female classmates. We hypothesize that girls perform better when a girl is top of the class in math rather than a boy. 811 French 7th graders (49% females) were asked to fill in a questionnaire. We collected their initial level of math performance. In each class we identified the gender of the student who is the best at math based on their average grades at the end of grade 7. To test our hypothesis, we performed a series of multilevel regressions to explain the level of math performance at the end of grade 7. The initial level of math performance was introduced as a control variable, we then introduced students’ gender, gender of the top of the class in math and teachers’ gender. Consistent with our hypothesis, in classes where a girl is top of the class in math, girls and boys perform equally. However, in classes where a boy is top of the class in math, girls significantly underperformed boys. These first results are promising. First, we will discuss them in terms of reducing ST. We will then suggest an experiment to corroborate these first results.