The human operator is a paradoxical element: in situation of stress or tiredness, it can be an element of the loss of the reliability of a system. However, in certain critical situations of insecurity, it can be a factor of reliability, by restoring the good performance of the system, sometimes by actions not envisaged by the safety regulations of the exploitation but dependent on its knowledge and its experiment. It is thus necessary to optimize the place of the man in the transportation system in full knowledge of its capacities but also of its limits. Research in this field proposed several of approaches and of models primarily of a conceptual nature: the model of Rasmussen, the model of Reason, the model of Van Eslande, the model of Neboit and the model of Leplat. In spite of the undeniable interest of these models, their implementation in the field of transport safety in particular railway is not satisfactory. No model alone makes it possible to ensure the exhaustiveness of the analysis. Indeed, these models are complementary, but none deals with the problem as a whole what proves the difficulty of their exploitation. No model, except that of Van Eslande, is supported by an industrial application or a real case of accidents. From where interested the objective of this article which attempts to bring a brief reply in order to make these model exploitable and applicable in the field of transport safety railway.