Urban rivers are often engineered to increase flood conveyance and stabilize channel size and position. This paper analyses habitat surveys of 180 urban river stretches of differing engineering type from four river basins (river Tame, West Midlands, UK; tributaries of the lower river Thames, UK; river Botic, Prague, Czech Republic; river Emscher, North-Rhine Westphalia, Germany). Kruskal–Wallis tests identify significant differences in extent and/or frequency of flow types, bank and bed physical habitats, and vegetation characteristics associated with different styles of engineering. Principal Components Analysis identifies four key environmental gradients in the data set: sediment supply and retention; extent and diversity of in-channel vegetation and riparian trees; bed and bank sediment calibre; flow type energy and complexity. These gradients discriminate stretches of differing planform, cross section and reinforcement and are significantly correlated with indices of degree and type of bank and bed reinforcement, pollution and presence of alien nuisance plant species. The analytical results illustrate statistically significant associations between different styles and levels of engineering intervention and the number and nature of physical habitats present in urban rivers. The results provide a basis for filtering sites for potential remedial measures prior to site-specific surveys and modelling, for comparing sites and for tracking trajectories of change at sites that are subject to changes in channel engineering. They provide evidence that river condition and degree of engineering are not inversely related in a simple linear way, and that engineering of urban river channels, in the form of mixed, patchy reinforcement can contribute a great deal to habitat diversity where other controls on flow heterogeneity are more difficult to manipulate. Copyright © 2011 John Wiley & Sons, Ltd.