AbstractI develop a structural model of urban travel to estimate long‐run gasoline price elasticities. I model the demand for transportation services using a dynamic discrete‐choice model with switching costs and estimate it using a panel dataset with public market‐level data on automobile and public transit use in Chicago. Long‐run own‐ (automobile) and cross‐ (transit) price elasticities are substantially more elastic than short‐run elasticities. Elasticity estimates from static and myopic models are downward biased. I use the estimated model to evaluate the response to several counterfactual policies. A gasoline tax is less regressive after accounting for the long‐run substitution behavior.