Today, as it is often difficult to demonstrate the superiority of a new molecule or a therapeutic strategy in term of plain efficacy on disease, the incitement is strong to provide some complementary argument of assessment, we are assisting to the emergence of a new concept: shared therapeutic decision making. Is the application of this concept – wich make the paternalistic model questionnable – adapted to all cases? What are the different levels of participation that could be envisaged? Are there favourable methods for this participation? This shared decision making – direct (patients’ choice between treatment options) or indirect (integration of elicited preferences in the decision process) – if it has to be efficient, must surround with care: to define its application limitation, to protect itself of manipulation. It shall require to consider information transmission difficulties, to establish some elicitation preference method. Some technical, such as time trade off, standard gamble or willingness to pay, supported by economic theory of expected utility, permit to help eliciting patients’ preferences and to structure the therapeutic choice. Some empirical study of preference elicitation shall permit to get clear the complexity of trade off between the different choice element that could enter in the acceptability of the treatment for patients.