Although free-roaming cats can impact the environment, and substantial resources have been invested to find humane ways to manage these cat populations, there are no empirical estimates of free-roaming cat population sizes in cities. Using Guelph as a case-study, we applied a replicated distance transect sampling and likelihood-based hierarchical modeling to compare human-mediated landscape patterns of land use, distance to roads, distance to wooded areas, building density, and socio-economic status to explain the abundance of free-roaming cats. We then derive an empirical estimate of total population size and present a spatially-explicit prediction of free-roaming cat density across the city. The prevalence of cats was highest in residential areas and lowest in commercial and institutional areas, negatively related to median household income, and positively related to distance from woods and building density. Total population size was estimated to be 7,662 free-roaming cats (95% bootstrap CI: 6,145 - 9,966) for Guelph; free-roaming cat density varied from 0 - 49.4 cats/ha. Our estimate overlapped with an independent estimate of indoor-outdoor cats (11,927; 95% CI: 6,361 - 20,989) derived from random surveys of city residents, which implies our methodology was relatively robust and unbiased. Our results suggest that free-roaming cat density in cities could be determined by bottom-up processes (e.g. enhanced food availability in residential areas) as well as top-down processes (e.g. enhanced susceptibility to coyote predation near wooded areas) which are typically reserved to explain wildlife populations in natural environments. These results guide ongoing local discussions on cat management, bylaws and urban planning.