This paper proposes a data-driven exploratory agent-based model representing a tobacco town. To investigate the effects of policy interventions on accessibility (cost and availability) and consequently, tobacco consumption, the model compares two economically diverse localities of the city of Glasgow namely Glasgow Kelvin (urban rich) and Glasgow Shettleston (urban poor). Several policies were simulated to reduce the density of retailers in the environment: These were (1) retailer density cap—a random reduction at a percentage of the original density (2) retailer and school proximity buffers at varying distances, and (3) a combination. The results showed that as retailer density decreased, overall costs increased with a threshold effect. For both localities of Glasgow, retailer buffers were predicted to be the most effective single policy to increase costs hence reduce accessibility. However, single policies were generally predicted to exhibit relatively mild effects on cost; a combination of policies contextualized to the locality was necessary to increase the cost of purchasing cigarettes to a degree that might alter smoking behavior. This study demonstrates the usefulness of ABMs in predicting the effects of public health policies and highlights the importance of considering the contextual effect of policies on smoking behaviors.