The human dimension of weed management is most evident when farmers make decisions contrary to science-based recommendations. Why do farmers resist adopting practices that will delay herbicide resistance, or seem to ignore new weed species or biotypes until it is too late? Weed scientists for the most part have ignored such questions or considered them beyond their domain and expertise, continuing to focus instead on fundamental weed science and technology. Recent pressing concerns about widespread failure of herbicide-based weed management and acceptability of emerging technologies necessitates a closer look at farmer decision making and the role of weed scientists in that process. Here we present a circular risk-analysis framework characterized by regular interaction with and input from farmers to inform both research and on-farm risk-management decisions. The framework utilizes mental models to probe the deeply held beliefs of farmers regarding weeds and weed management. A mental model is a complex, often hidden web of perceptions and attitudes that govern how we understand and respond to the world. One's mental model may limit ability to develop new insights and adopt new ways of management, and is best assessed through structured, open-ended interviews that enable the investigator to exhaust the subjects inherent to a particular risk. Our assessment of farmer mental models demonstrated the fundamental attribution error whereby farmers attributed problems with weed management primarily to factors outside of their control, such as uncontrolled weed growth on neighboring properties and environmental factors. Farmers also identified specific processes that contribute to weed problems that were not identified by experts; specifically, the importance of floods and faulty herbicide applications in the spread of weeds. Conventional farmers expressed an overwhelming preference for controlling weeds with herbicides, a preference that was reinforced by their extreme dislike for weeds. These preferences reflect a typical inverse relationship between perceived risk and benefit, where an activity or entity we perceive as beneficial is by default perceived as low risk. This preference diminishes the ability of farmers to appreciate the risks associated with overreliance on herbicides. Likewise, conventional farmers saw great risk and little benefit in preventive measures for weed control. We expect that thorough two-way communication and a deeper understanding of farmer belief systems will facilitate the development of audience-specific outreach programs with an enhanced probability of affecting better weed management decisions. © 2010 Weed Science Society of America.