This paper exploits longitudinal employer-employee matched data from the U.S. Census Bureau to investigate the contribution of worker and firm reallocation to changes in earnings inequality within and across industries between 1992 and 2003. We find that factors that cannot be measured using standard cross-sectional data, including the entry and exit of firms and the sorting of workers across firms, are important sources of changes in earnings distributions over time. Our results also suggest that the dynamics driving changes in earnings inequality are heterogeneous across industries.