In the debate on urban inequality, Sassen’s theory on social polarization and Wilson’s theory on spatial mismatch have received much attention. Where Sassen highlights the decline of the middle classes, Wilson focuses on the upgrading of urban labour markets. In this article we argue that both theories may be valid, but that they have to be put in a more extended theoretical framework. Of central importance are national institutional arrangements, membership of different ethnic groups and networks, and place–specific characteristics rooted in local socio–economic histories. As a first empirical illustration of our model, we use data on the labour markets of Amsterdam and Rotterdam and show that different forms of inequality can be found both in economic sectors and within ethnic groups. The model we present could be used both to reinterébatpret existing data and as an analytical framework for the analysis of different forms of urban inequality.