A real time engine model to be used as an onboard observer in Large Civil Turbofans has been developed. The main features of this Real Time Model are its non-linear physical structure (identical with a detailed transient model, which may represent the real engine) as well as its ability to adapt itself to the engine condition. General and specific requirements for the targeted model have been set, evolution of the model is presented and various investigations concerning the ability of the model to satisfy these requirements are discussed. The conclusion is that the model is faster than real time in terms of a realistic time step for current turbofan engines controllers using a processor with computing power about ten times less than the current existing processors. It is also worth mentioned that this speed is achieved without sacrificing model accuracy. In general maximum errors less than 1% can be obtained using no more than five fixed iterations in the model in order to achieve this speed.