Authorities define cities-or human settlements in general-through imposing top-down rules in terms of whether buildings belong to cities. Emerging geospatial big data makes it possible to define cities from the bottom up, i.e., buildings determine themselves whether they belong to a city using the notion of natural cities and based on head/tail breaks, which is a classification and visualization tool for data with a heavy-tailed distribution. In this paper, we used 125 million building locations-all building footprints of America (mainland) or their centroids more precisely-to generate 2.1 million natural cities in the country (see the URL as shown in the note of Figure 1). In contrast to government defined city boundaries, these natural cities constitute a valuable data source for city-related research.