Income differences across countries are large. The literature has yet to settle on what share of those differences can be attributed to human capital, as well as on what determines differences in human capital. This paper contributes to both questions. It uses unique Swedish employer-employee linked administrative data to estimate differences in human capital as country-of-origin specific labor productivity terms in firm production functions. Unlike previous migrant-based measures in the literature, the one in this paper is immune to concerns related to wage discrimination and robust to other varieties of discrimination. After accounting for education and experience, estimated human capital still varies by a factor of 3 between the countries at the 90th and 10th percentile of the human-capital distribution. When I investigate which country-of-origin characteristics most closely correlate with my estimates of human capital, cultural values elicited from the World Values Survey are the only robust predictor. This relationship also persists among the children of migrants, which lends further credence to the cultural interpretation of human-capital differences unexplained by education and experience. Finally, adjusting human-capital stocks by the estimated cultural component in a development accounting exercise decreases the amount of unexplained variation in income by 16 percentage points relative to “traditional” development accounting.