Abstract: We study gender differences in levels of school value-added and their role in explaining the math gender gap in Chile. Using rich and representative panel data on students’ test scores, we fit two-way (student and school) fixed effect models for each gender. We interpret school fixed effects as a value-added measure. We find that schools on average do not play a role in the determination of the math gender gap. However, there is heterogeneity in this effect with private schools helping to close this gap by 38.6%. In studying the mechanisms, we rule out the possibility that this average result for private schools is driven by girls being overrepresented in high value-added schools. The effect found is explained by a gender-specific effect: at a given private school, more value-added is obtained by girls relative to boys. Finally, we find that the presence of a higher share of female teachers and attending a private school correlates with higher returns to girls’ value-added, and that average teachers’ expectations regarding students’ future outcomes play a more favorable role for boys.