Agricultural production decision-making is becoming more complex. This is due, in part, to the increased competition caused by the globalization of agriculture, the need to adopt more sustainable farming practices and the increasing rate and volume of information exchange. Predicting the behavior of production systems is critically important for sustainability issues that concern profitability, environment soundness and consistency with society interests. This raises an important systems research demand for innovation in agricultural production management. Computer based simulation is one of the most commonly used tools utilized to aid in the design and evaluation of production management policies. Simulation approaches have traditionally focused on the agronomic and technological aspects of the production processes, e.g. crop or animal responses to farming operations. There is a growing recognition by the scientific community that principles from the biological, physical, management and social sciences must be integrated to understand and predict the behavior of agro-ecosystems under the various dimensions that are relevant for design purpose. The prevailing simulation models used by the agronomists and farming systems researchers largely abstract from the spatial, temporal and resource boundedness aspects of agricultural production systems. Consequently they offer limited help in explaining system performances and their sensitivity to external events and management behavior. How then can we develop a more useful simulation approach to understand and predict the behavior of agricultural production systems? Using biophysical models based on scientific knowledge of soil, crop and animal functioning is definitely necessary but may not be sufficient. The paper argues in favor of a more realistic, explicit, systemic representation of the production management behavior that includes the following features: • decision-making is driven by a management policy that is specific to the farmer's material conditions and constraints, his understanding or belief about the biophysical functioning of his system and his personal expertise about relevant decision indicators; • since the technical operations usually interact (utilize the same resource pool and have synergic or conflicting effects), they should not be dealt with separately; • the decision-making process is both reac- tive (response to endogenous and exoge- nous events) and anticipatory (goal- oriented); • decision-making deals primarily with the issue of dynamic work scheduling and resource allocation; • the simulated decision process must reflect the way uncertainty and multiple goals are dealt with to select actions in situ. Basically we advocate for a modeling and simulation approach that recognizes the central role of humans in the functioning of agricultural production systems. Besides the biophysical aspect, the model should pay particular attention to the manager's decision making process and to the aspects of implementation of the actions decided. After having recalled the rationale behind this study, the paper reviews the notion of agricultural production management and outlines the objective and scope of the advocated approach. The main section presents the salient features that the model should possess. Finally some difficulties that might be encountered with the practical use of such an approach are pointed out.