Although the gender disparity in favor of men in the overall education attainment has declined or even reversed in many countries, the underrepresentation of women in science, technology, engineering, and math (STEM) fields persists across the world. Using cross-national education survey data from the 2015 Programme for International Student Assessment (PISA), this dissertation provides in-depth analysis on the role of social structure in shaping gender stratification in STEM education. In particular, it examines how national, school, and family factors interact to shape gender differences in STEM performance, attitudes, and aspirations of 15-year-old students. Chapter 1 examines whether greater national gender equality is associated with improved math performance for girls relative to boys. To answer this question, it exhibits gender disparity in math performance across the distribution and examines how gender gaps vary according to various domains of national-level gender equality regimes. Results show that greater gender wage equality is associated with smaller gender gaps across the performance distribution, while greater female representation in professional and technical jobs and a larger women’s share in the service sector are associated with smaller gender gaps among top math performers. Findings provide evidence to the importance of opportunity structures in shaping gender disparities in math performance, and the possibility for improved female performance where females are better represented and rewarded in STEM fields. Chapter 2 examines processes behind the “paradox of affluence”, or the phenomenon that male advantages in STEM attitudes and aspirations tend to be larger among students in more affluent countries and families. Three processes are proposed to understand this paradox: gendering of intrinsic interest in math, gendering of instrumental motivation to learn math, and gendering of science self-efficacy in more affluent settings. Further, results show that the gendering effect of family affluence can be reinforced by national affluence. This study thus highlights the interaction of national and family contexts in shaping gender differences in STEM attitudes and aspirations. Chapter 3 explores whether national education system characteristics and school characteristics are associated with gender differences in science attitudes. Findings show that gender differences in student science interest and motivation are larger in more standardized, stratified, and privatized education systems; conversely, the gender gap in science self-efficacy is smaller in such systems. Further, school autonomy is found to have a gendering effect on student science attitudes, and this effect is even stronger in more standardized education systems. Results underscore the complexities of education institution structures in shaping gender differences in STEM fields and provide insights to address gender disparity in STEM attitudes.