Evolutions of the institutional and environmental contexts are driving search for alternative cropping systems to reduce water use while maintaining high levels of productivity. This thesis is an account of the long tradition of research on cropping-plan choices at the farm level. It concerns the scope of modelling agricultural systems with an opening to economy. The objective of the research described in this thesis is to produce formalised knowledge on farmers’ croppingplan choices under uncertain environment (price and weather) by analysing and modelling their decision-making processes. Formalising and modelling decision-making processes is becoming a crucial point to develop decision support systems that go beyond limitations of formerly developed prescriptive approaches. This thesis contributes to the development of a formalised and integrated methodology to study and model complex decision-making process. This methodology enables to fill the gap between field surveys and decision-model implementation. The methodology is drawn upon a theoretical background of the decision-making, and consistently combined tools to respectively survey, analyse, model and implement coupled agent and biophysical models. In this thesis, I address the question of uncertainty in two directions. I first analyse the spatio-temporal dynamic of individual farmers’ decision-making process. Then I estimate farmers’ aversion to risk by comparing stated and revealed elicitation methods. On the basis of field survey results, I develop a decision model called CRASH. The approach to develop the model stresses on explicit formalisation of the decision-making process in its temporal and spatial dimensions, and representation of the domain knowledge through generic concepts that are close to ones used by decision-makers. The implementation of developed models is carried out on the RECORD platform as trail blazing project. The originality relies on the use of dynamic models on both the biophysical and management processes. This research opens new perspectives for developing farm specific decision support systems that are based on simulating farmers’ decisionmaking processes. The modelling and simulating the cropping-plan decision-making processes should enable to design with farmers cropping systems that re-conciliate the required adaptive capacities and needs to maintain cropping systems productivity.