Urbanization affects key aspects of wildlife ecology. Dispersal in urban wildlife species may be impacted by geographical barriers but also by a species’ inherent behavioural variability. There are no functional connectivity analyses using continuous individual‐based sampling across an urban‐rural continuum that would allow a thorough assessment of the relative importance of physical and behavioural dispersal barriers. We used 16 microsatellite loci to genotype 374 red foxes (Vulpes vulpes) from the city of Berlin and surrounding rural regions in Brandenburg in order to study genetic structure and dispersal behaviour of a mobile carnivore across the urban‐rural landscape. We assessed functional connectivity by applying an individual‐based landscape genetic optimization procedure. Three commonly used genetic distance measures yielded different model selection results, with only the results of an eigenvector‐based multivariate analysis reasonably explaining genetic differentiation patterns. Genetic clustering methods and landscape resistance modelling supported the presence of an urban population with reduced dispersal across the city border. Artificial structures (railways, motorways) served as main dispersal corridors within the cityscape, yet urban foxes avoided densely built‐up areas. We show that despite their ubiquitous presence in urban areas, their mobility and behavioural plasticity, foxes were affected in their dispersal by anthropogenic presence. Distinguishing between man‐made structures and sites of human activity, rather than between natural and artificial structures, is thus essential for better understanding urban fox dispersal. This differentiation may also help to understand dispersal of other urban wildlife and to predict how behaviour can shape population genetic structure beyond physical barriers.