Rainfed lowland rice production in lower northeast Thailand is a complex and adaptive farming activity. Complexity arises from interconnections between multiple and intertwined processes, particularly agronomic practices and labor migration. Having faced a heterogeneous and very variable environment for centuries, local rice farmers are very adaptive and used to adjust their behavior in unpredictable farming conditions. Based on the principles of the Companion Modelling (ComMod) approach, indigenous and academic knowledge was integrated in an Agent-Based Model (ABM) to build a shared representation of this complex and adaptive system. The ABM was codesigned over a 3 years long process with a group of different types of rice growers from Ban Mak Mai village in southern Ubon Ratchathani province. This participatory modeling process also aimed at stimulating participants¿ thinking and co-learning through the collective exploration of simulated scenarios with varying levels of water and labor availability, with the ultimate goal of further strengthening their adaptive management ability. The ABM consists of three interacting modules in a virtual rainfed lowland rice environment: Water (hydro-climatic processes), Rice, and Household. "Household" is a rule-based agent making daily decisions based on its available means of production depending on the stage of the rice crop, and water and labor availability. Key decisions made are related to: i) rice nursery establishment, ii) rice transplanting, iii) rice harvesting, and iv) migration of household members. The spatially-explicit model interface represents an archetypical toposequence made of upper to lower paddies in a minicatchment farmed by 4 different households and includes also water bodies and human settlements. According to participating farmers, after many iterations between the lab. and the field, this ABM adequately represents their rice farming and labor migration management practices. Using the model to raise f