This paper investigates whether individuals living in neighbourhoods with high concentrations of unemployment are less likely to enter work if they are unemployed and more likely to lose their job if they are employed. The main challenge in the neighbourhood effects literature is the identification of causal neighbourhood effects. A particular problem is that individuals do not randomly select neighbourhoods to live in: the selection process is often linked to the labour market situation and potential of individuals. To get more insight in neighbourhood effects we run separate models for social renters and owner occupiers. This study uses anonymised individual level longitudinal data from the Scottish Longitudinal Study for 1991 and 2001 with multiple neighbourhood scales operationalised. Based on the results we argue that any apparent neighbourhoods effects that were present in models of the full population are at least partly an artefact of different neighbourhood selection mechanisms. The conclusions of the paper call for a more nuanced treatment of neighbourhood effects and the development of models that seek to include neighbourhood selection from the outset.