Despite the regularity of disasters, social science has only begun to generate replicable knowledge about the factors which facilitate post-crisis recovery. Building on the broad variation in recovery rates within disaster-affected cities, I investigate the ability of Kobe's nine wards to repopulate after the 1995 Kobe earthquake in Japan. This article uses case studies of neighborhoods in Kobe alongside new time-series, cross-sectional data set to test five variables thought to influence recovery along with the relatively untested factor of social capital. Controlling for damage, population density, economic conditions, inequality and other variables thought important in past research, social capital proves to be the strongest and most robust predictor of population recovery after catastrophe. This has important implications both for public policies focused on reconstruction and for social science more generally.Keywords Kobe earthquake Á Population recovery Á Social capital Á Disaster Á Resilience Disasters remain among the most critical shocks which impact residents and their neighborhoods around the world. While large-scale crises such as the Indian Ocean tsunami (2004), Hurricane Katrina (2005), and Italy's earthquake (2009) captured media attention, numerous smaller-scale floods, typhoons, earthquakes, and mudslides killed hundreds of thousands of victims around the world and affected far more. Researchers have confirmed an upward trend in the number in the number of disasters, individuals affected by them, and their economic costs over the past two decades (Hoyois et al. 2007). A better understanding of disasters, or more specifically of disaster recovery, will provide policy makers and victims alike with ''usable knowledge''-knowledge that is both accurate and politically tractable (Haas 2004: 572).