Studies in the land use–travel connection have long neglected the role of neighbourhood crime. This paper analyses such a role of crime and explores whether more compact land-use characteristics tend to cluster more crimes, thus having a negative impact on public transit use. A path analysis model is used to estimate the relationship between land use, crime, and bus ridership in the city of Austin, Texas. The results demonstrate that higher population density and more mixed land use may significantly stimulate more crimes near the bus stops while the level of crime may have a nonlinear effect on ridership. Once the number of reported crimes exceeds a threshold level, ridership is negatively affected. As a result, very dense residential and commercial development may trigger a rise in crimes, leading to a fall in ridership. These findings suggest that those who seek to implement land use policies to increase transit use and reduce auto dependence should recognise and evaluate land use impact on crime and how this impact affects travel behaviours.