We revisit the role of human capital in explaining the cross-country variation in GDP. We propose a general-equilibrium model in which workers of different human- capital groups (education and experience) sort across broad occupational categories. We map the model to a unique harmonized micro dataset that allows to measure average wages by human capital and occupation for 50 countries that span the entire development spectrum. The calibration reveals that rich countries have particularly high productivity in more complex, white-collar occupations. The composition and quality of human capital explain half of the cross-country non agricultural GDP per-worker gap relative to the US. For the poorest quintile of countries, a shift to US human capital would double non-agricultural GDP and the white-collar employment rate. We also find that occupational distortions are more pronounced in poor countries. They depress white-collar employment and contribute to a high white-collar wage premium.