There is a growing interest in exploring the relationships between the built environment and auto ownership and a number of studies have investigated the impact of rail transit on travel behaviour. However, few have disentangled the impact of rail transit on auto ownership from the influences of the built environment and residential self-selection. Using the light rail transit (LRT) in the Minneapolis-St. Paul metropolitan area, USA, this study applies the statistical control approach and quasi-longitudinal design to examine the effects of LRT, neighbourhood design and self-selection on auto ownership. It is found that residential self-selection influences auto ownership; backyard size, off-street parking and business density marginally affect auto ownership; and the LRT does not have an independent impact on auto ownership beyond neighbourhood design and self-selection. The results point to the importance of neighbourhood design in rail transit development.