The monocentric city model is generalized to a fully structural form with leisure in utility, congested commuting, and the equalizing of utility and perimeter land price across metros. Exogenous and agglomerative differences in total factor productivity (TFP) drive differences in metro population, radius, land use, commute time, and home prices. Quantitative results approximate observed correspondences among these outcomes across U.S. metros. Traffic congestion proves the critical force constraining population. Self-driving cars significantly increase the sensitivity of metro population to productivity. Population becomes less responsive to increases in productivity as metros become larger. Correspondingly, the productivity “cost” of metro population—the TFP required to support a given population—increases convexly with size. Benchmark estimates suggest that agglomerative productivity suffices to support increases in population from low levels, allowing chance to play a significant role in determining which locations with sufficient exogenous TFP develop into small metros. But agglomerative productivity falls considerably short of supporting increases in population from high levels, suggesting that large metros arise from strong “fundamentals” such as high exogenous TFP.