Natural disasters are known to have devastating immediate impacts, but their long-run effect on economic growth is not well understood. For the natural hazard of earthquakes, this paper provides the first global empirical study on this topic that applies a measure of the exogenous physical hazard responsible for earthquake impacts, earthquake ground shaking. I exploit the random within-country year-to-year variation of shaking to identify the causal effect of earthquakes on economic growth. To construct a panel dataset with country-year observations of earthquake exposure and socioeconomic variables, I combine the universe of relevant earthquake ground shaking data from 1973 to 2015 with country-level World Bank indicators. I find negative long-run growth impacts for an average country comparable with recent findings for climate related natural disasters. A typical earthquake reduces GDP per capita by 1.6% eight years later, with substantial heterogeneity by country categories. In particular, low and middleincome countries experience the greatest long-run economic damages while high-income countries may even experience some positive "building back better" effects. Based on an analysis of alternative spatial aggregation approaches, I find earthquake impacts are driven by local high-intensity events rather than spatially diffuse exposure to lower intensity shaking.