Abstract Manufacturing systems are faced with the increasing constraints of productivity and flexibility. The problems to be solved are related to the control of the shop-floor (scheduling, dispatching, activity control), and the equipment (monitoring, maintenance, reliability, set-up). Materials management must be defined in order to ensure the synchronization and coordination of the different resources and lead to a ‘just-in-time’ approach. The hierarchical control structures previously described cannot provide complete and fully efficient solutions. A dynamic and distributed structure for flexible manufacturing cells (FMC) derived from AI techniques for distributed problem-solving is presented. This structure is based on the negotiation approach. An entity of the manufacturing system—workstation, cell, production line, handling system—is considered as an autonomous agent, able to co-operate with other agents to achieve a production program. Co-operation is performed by exchanging messages between the different agents. Different models of the control system are presented: the conceptual model, associated to an agent; the integrated management station (IMS), which has to carry out local management and communication functions (it includes a multi-model decision system involving algorithms and heuristics, an information system and a control system); the logical model, built with Petri Nets, which defines the type of messages and the information flow; and the simulation model, using an object-oriented approach.