Rangeland-based livestock systems have to deal with the significant instability and uncertainty of the agricultural context (policy changes, volatility of input prices, etc.), and especially of the climatic context. Thus, they are particularly concerned by adaptive management strategies. To support the development of such strategies, we developed a board game including a computer model called "Rangeland Rummy". It is to be used by groups of farmers and agricultural consultants in the context of short workshops (about 3 h). Rangeland Rummy builds upon five types of material object: (i) a game board; (ii) a calendar stick indicating the starting date of the game board; (iii) sticks marked with the feed resources available for combinations of vegetation types and their management practices; (iv) cards to define animal groups and their feeding requirements throughout the year; (v) cards related to types of feed that can be attributed to animal groups throughout the year. Using these material objects, farmers collectively design a rangeland-based livestock system. This system is immediately evaluated using a computer model, i.e. a spreadsheet providing graphs and indicators providing information on, among other things, the extent to which quantitative and qualitative animal feeding requirements are covered across the year. Playing the game thus consists in collectively and iteratively designing and evaluating rangeland-based livestock systems, while confronting the players with new contextual challenges (e.g. interannual variability of weather, volatility of input prices) or new farmers' objectives (e.g. being self-sufficient for animal feeding). An example of application of Rangeland Rummy with 3 farmers in southern France is reported. Applications show that it tends to develop farmers' adaptive capacity by stimulating their discussions and the exchange of locally-relevant knowledge on management strategies and practices in rangeland-based livestock systems.