In this paper, we review the evidence base for social mixing in neighbourhoods, which is used as a strategy to tackle assumed negative neighbourhood effects. We discuss in detail the theoretical links between neighbourhood characteristics, and outcomes of individuals living in concentrations of poverty. Through this we identify the theoretical case for promoting socially mixed communities. We then review the empirical evidence base, focusing on outcomes of the American poverty deconcentration initiatives including the Moving to Opportunity and HOPE VI programs. We identify that the evidence from these programs is at best inconclusive. Turning to the European experience we identify problems associated with using observational data to assess individual outcomes in relation to their neighbourhood context. We conclude by suggesting that the evidence base for social mixing is far from robust, and that many of the current empirical papers suffer from serious analytical shortcomings. Ultimately, the process of creating more socially mixed neighbourhoods is unlikely to create more opportunities in life for the original residents. Socially mixing neighbourhoods through tenure mixing will only change the population composition of neighbourhoods, increasing average incomes because more affluent (and employed) residents will move into the owner occupied housing replacing social housing.