Based on a literature review, an international benchmark, and a case study in Ontario, the dissertation explores farmers' preferences for agri-environmental incentive programs. We use literature in agricultural, environmental and behavioural economics, and social psychology to propose an original analytical framework of farmers' preferences. This framework is tested in Ontario and helps understand farmers' motivations to participate or not to participate in environmental incentive programs. We then review the evolution of agri-environmental programs in Ontario since the 1930s. We develop original frameworks to compare programs fostering either collaboration or competition among farmers and use these frameworks to compare Ontario programs with contrasting programs in other provinces and OECD countries. Based on program design analysis and semi-structured interviews, we identify program characteristics that matter to Ontario farmers. Through a choice experiment, we then test which characteristics matter most in farmers' decisions and segment Ontario farmers into groups of heterogeneous preferences. The dissertation decisively adopts a mixed-methodology approach. Qualitative research is used to design the choice experiment (exploratory sequential design). Open-ended questions in the experiment link farmer choice-profiles with qualitative results, thus confirming, illustrating or nuancing choice experiment findings (convergent design). Finally, the participation of informal field advisors all along the research process allows for a research design targeted on current policy questions and continuous verification of intermediate findings. We find that programs designed to be the most competitive are not necessarily the most cost-efficient because medium and long-term dimensions are not considered. We identify five farmer profiles with different preferences and attitudinal traits. While 'Business Farmers' and 'Busy Farmers' have a strong preference for high incentive levels, other types of farmers either value technical assistance above the monetary incentive, are only willing to participate if their contact point is a fellow farmer, or are largely unwilling to participate. Several farmer profiles would prefer collective approaches. We also find that preferences are endogenous. They evolve as information emerges, applications succeed or fail, engagement takes place, and farmers discuss approaches and their impacts. The concluding chapter presents, on this basis, a series of research-grounded recommendations for agri-environmental policy design in Ontario.