Population health differs dramatically across neighbourhoods. Contextual determinants of health such as neighbourhood effects have become a focal point of research. However, the scientific community is still actively debating whether neighbourhood effects are causal (i.e. living in a certain neighbourhood causally affects one’s health) or merely the result of selection effects (i.e. healthier people moving in an out of certain neighbourhoods). This distinction is crucial for policymakers, as only the identification of causal effects could inform policies and intervention initiatives. Recent literature argued that the process that selects people into neighbourhoods has a theoretical meaning and should be explicitly studied and modelled. This can be done by implementing a two-step approach based on Heckman’s two-stage regression framework: 1) modelling the neighbourhood choice of people that relocate between neighbourhoods, and 2) incorporating these results in a model that investigates the effect of the neighbourhood on the outcome of interest (in this case, health). The Microdata and computing facilities of Statistics Netherlands provide a unique opportunity to adopted these methods and advance our understanding of the effects of neighbourhoods on health outcomes. The aim of this project is to estimate the effect of neighbourhood characteristics (e.g. neighbourhood deprivation, demographic mix and safety) on health behaviours (smoking and alcohol use) and medicine use (as an indicator of health and mental wellbeing). We will focus our research on the city of Rotterdam as this city has a wide variety in neighbourhood characteristics and provides unique opportunities for local policy evaluation. The proposed research project promises to contribute valuable information about the health effects of neighbourhoods, providing urgently needed information to policy-makers on opportunities to reduce the prevalence of smoking, alcohol use and mental health problems.