It is unclear whether policies designed to reduce auto and increase transit usage achieve their objective. Evidence is mixed because most empirical research on these policies use ad hoc specifications, whereas our models are drawn from economic theory. Three models of increasing generality show how endogenizing relevant variables changes results obtained by others. The theoretical hypotheses are empirically tested using a dataset that integrates travel and land use. Our main findings are (1) population density has a small impact on transit demand, which decreases when residential location is endogenous; (2) households living farther from work use less transit, a result of trip-chaining; and (3) reducing the spatial allocation of non-work activities, improving transit accessibility at and around subcenters, and increasing the presence of retail locations in proximity to transit-oriented households would increase transit demand.