Social frontiers arise when there are sharp differences in the demographic composition of adjacent communities. This paper provides the first quantitative study of their impact on household mobility. We hypothesise that conflicting forces of white flight and territorial allegiance lead to asymmetrical effects, impacting residents on one side of the frontier more than the other due to differences in the range of housing options available to different groups, and different symbolic interpretations of the frontier. Using Dutch registry data for the city of Rotterdam we identify ethnic social frontier locations using a Bayesian spatial model (Dean et al., 2019), exploiting the data’s one hundred metre resolution to estimate frontiers at a very small spatial scale. Regression analysis of moving decisions finds that the ethnic asymmetry of the frontier matters more than ethnicity of individual households. On the ethnic minority side of the frontier, households of all ethnicities in the 28–37 age range have reduced probability of moving compared to non-frontier parts of the city. The opposite is true on the Dutch native side of the frontier. We supplement this analysis with flow models which again find strong frontier effects. Our findings illustrate how the study of social frontiers can shed light on local population dynamics and neighbourhood change.