<p>In the USA, employment subcenters are widely recognized features of contemporary metropolitan regions. Yet despite the prominence of polycentric cities in theories of urban economics and a significant body of work measuring subcenters, there is remarkably little research devoted to understanding the distribution of subcenter sizes within metropolitan regions. This study applies Zipf’s law to the size distribution of subcenters within 78 metropolitan regions in the USA. For many of these regions, Zipf’s law provides a good basis for understanding the size distribution of subcenters, meaning that these regions contain a few large subcenters, and many smaller subcenters. These findings are robust across different methods of subcenter identification. The paper considers why Zipf’s law appears to fit the distribution of subcenter sizes better than it does the distribution of city sizes in the USA.</p>