This paper provides a framework for assessing empirical patterns of urban disaster recovery through the use of statistical indicators. Such a framework is needed to develop systematic knowledge on how cities recover from disasters. The proposed framework addresses such issues as defining recovery, filtering out exogenous influences unrelated to the disaster, and making comparisons across disparate areas or events. It is applied to document how Kobe City, Japan, recovered from the catastrophic 1995 earthquake. Findings indicate that while aggregate population regained pre-disaster levels in ten years, population had shifted away from the older urban core. Economic recovery was characterised by a three to four year temporary boost in reconstruction activities, followed by settlement at a level some ten per cent below pre-disaster levels. Other long-term effects included substantial losses of port activity and sectoral shifts toward services and large businesses. These patterns of change and disparity generally accelerated pre-disaster trends.