Recent empirical work in the obesity literature has highlighted the role of the built environment and its potential influence in the increasing prevalence of obesity in adults and children. One feature of the built environment that has gained increasing attention is the role of access to chain grocers and their impact on body mass index (BMI). The assessment of the impacts of spatial access to chain grocers on BMI is complicated by two empirical regularities in the data. There is evidence that health outcomes such as BMI are clustered in space and that there is spatial dependence across individuals. In this article, we use an econometric model that takes into account the spatial dependence, and we allow the effect of access to differ for a person depending on whether he or she lives in a low-income community or peer group. We categorize this community using the characteristics of the people who immediately surround the individual rather than using census tracts. Using georeferenced survey data on adults in Marion County, Indiana, we find that the effect of improvements in chain grocer access on BMI varies depending on community characteristics.