There is conflicting evidence on the consequences of immigrant neighbourhood segregation for individual outcomes, with various studies finding positive, negative or insubstantial effects. In this paper, we document the evolution of immigrant segregation in England over the last 40 years. We show that standard measures of segregation point to gentle declines over time for all immigrant groups. However, this hides a significant increase in the number of immigrant enclaves where immigrants account for a substantial fraction of the local population. We then explore the link between immigrant segregation, enclaves and crime using both recorded crime and self-reported crime victimization data. Controlling for a rich set of observables, we find that crime is substantially lower in those neighbourhoods with sizeable immigrant population shares. The effect is non-linear and only becomes significant in enclaves. It is present for both natives and immigrants living in such neighbourhoods. Considering different crime types, the evidence suggests that such neighbourhoods benefit from a reduction in more minor, non-violent crimes. We discuss possible mechanisms for the results we observe.