Fire-prone landscapes are not well studied as coupled human and natural systems (CHANS) and present many challenges for understanding and promoting adaptive behaviors and institutions. Here, we explore how heterogeneity, feedbacks, and external drivers in this type of natural hazard system can lead to complexity and can limit the development of more adaptive approaches to policy and management. Institutions and social networks can counter these limitations and promote adaptation. We also develop a conceptual model that includes a robust characterization of social subsystems for a fire-prone landscape in Oregon and describe how we are building an agent-based model to promote understanding of this social-ecological system. Our agent-based model, which incorporates existing ecological models of vegetation and fire and is based on empirical studies of landowner decision-making, will be used to explore alternative management and fire scenarios with land managers and various public entities. We expect that the development of CHANS frameworks and the application of a simulation model in a collaborative setting will facilitate the development of more effective policies and practices for fire-prone landscapes.