Transit in the United States often suffers from the problem of inability to deliver travelers all the way from their point of origin to their destination. This “last-mile” problem is thought to deter transit use among riders with auto access, even when high-quality transit service is provided for the majority of the trip distance. We developed an agent-based model representing the commuters and their preferences for different aspects of transportation disutility, namely cost, time and safety. Commuters in the model assess their transportation options in light of their preferences, the characteristics of their environment, and the various modes available to them. The model is calibrated with data from four Chicago neighborhoods, representing four different combinations of land-use patterns and household income. We use this model to examine how transportation improvements, including automated driverless shuttles between origins of trips and nearby transit stations, and physical improvements enhancing pedestrians’ and cyclists’ commute might help overcome the last-mile problem particularly as they interact with policy shifts including changing in parking and fuel costs.