Distributed control architectures in manufacturing are becoming increasingly popular because their modularity makes them easy to install, configure, and modify. These benefits do not come for free. For creating the manufacturing systems of the future, engineers need to dare a leap in their ways of thinking. Considering a simple multi-agent model in use in a distributed flexible manufacturing project, it is shown that the dynamics of such multi-agent systems without appropriate design control can be chaotic, strongly dependent on some parameters inherent to the agent's negotiation rules. In order to develop appropriate multi-agent models, the dynamic behavior of the system is investigated using concepts and methods of the theory of nonlinear dynamical systems, and suggests some ways to control chaotic behaviours.