Using PISA test scores from 11,527 second-generation immigrants coming from 35 different countries of ancestry and living in 9 host countries, we find that the positive effects of country-of-ancestry gender social norms on girls' math test scores relative to those of boys: (1) expand to other subjects (namely reading and science); (2) are shaped by beliefs on women's political empowerment and economic opportunity; and (3) are driven by parents' influencing their children's (especially their girls') preferences. Our evidence further suggest that these findings are driven by cognitive skills, suggesting that social gender norms affect parent's expectations on girls' academic knowledge relative to that of boys, but not on other attributes for success--such as non-cognitive skills. Taken together, our results highlight the relevance of general (as opposed to math-specific) gender stereotypes on the math gender gap, and suggest that parents' gender social norms shape youth's test scores by transmitting preferences for cognitive skills.