This thesis concerns the hybrid traffic flow modelling and simulation. This approach uses jointly the two classical traffic flow models ; macroscopic and microscopic. It is a matter of adapting the representation to the studied phenomenon. Indeed, the macroscopic model is well adapted to the motorway network representation whereas the microscopic one makes it possible to describe an urban network more finely. Thus, we can obtain detailed information of the traffic state on a crossroads (for example) using the microscopic model and to implement traffic control laws synthesized from a macroscopic one. The final objective is to design and implement a decision-making tool allowing us to test and validate various traffic control strategies. For that, we initially developed a microscopic model founded on the paradigm agent. Then, we have proposed a coupling procedure, which makes it possible to ensure the conservation and the continuity of flow through the transition zone that connect the two worlds ; microscopic and macroscopic. The simulation results of the developed hybrid model show that it is possible to cohabit two models (microscopic and macroscopic) within the same system. To illustrate the relevance of the hybrid model, a traffic flow control application has been proposed. The results show that the vehicles travel total time was reduced. The proposed model represents a first version of a traffic flow hybrid simulator, which will remain to be developed within the furthers works.