We quantify the causal effect of schooling on cognitive skills across 21 countries and the full distribution of working-age individuals. We exploit exogenous variation in educational attainment induced by a broad set of institutional reforms affecting different cohorts of individuals in different countries. We find a positive effect of an additional year of schooling on internationally-comparable numeracy and literacy scores. We show that the effect is substantially homogeneous by gender and socio-economic background and that it is larger for individuals completing a formal qualification rather than dropping out. Results suggest that early and late school years are the most decisive for cognitive skill development. Exploiting unique survey data on the use of skills, we find suggestive evidence that our result is mediated by access to high-skill jobs.