Despite the recent flurry of interest in various aspects of ancient urbanism, we still know little about how much traffic flowed in and out of ancient cities, in part because of problems with using commodities as proxies for trade. This article investigates another approach, which is to estimate these flows from the built environment, concentrating on transport infrastructure such as city gates. To do this, I begin by discussing a new model for how we would expect this kind of infrastructure to expand with population, before investigating the relationship between the populations of sites and the total numbers and widths of city gates, focusing on the Greek and Roman world. The results suggest that there is indeed a systematic relationship between the estimated populations of cities and transport infrastructure, which is entirely consistent with broader theoretical and empirical expectations. This gives us a new way of exploring the connectivity and integration of ancient cities, contributing to a growing body of general theory about how settlements operate across space and time.