The relationships between franchisors and franchisees are more dependent on relational and psychological contracts, emerging in a day to day life of the franchise, than from the legal contract. But a few studies are devoted to the organisation of decisions between franchisors and franchisee, or franchisee autonomy. To better understand these relations, we studied the franchise context and existing literature. Then we performed a qualitative analysis among franchise experts, franchisors and franchisees. We carried out semi-directive interviews with 13 franchisees, 5 franchisors and 11 franchise experts and, after a first phase of reading with a floating attention, we performed a lexical analysis, with Alceste software, breaking down the corpus based on interviewee profile (franchise experts, franchisors and franchisees). A third phase, a thematic analysis, consisted in manual coding assisted by N’Vivo. We were able distinguish and define the concepts of independence and autonomy, sometimes blurred in the literature, and to develop a theoretical model, a causal paradigm one. The aim of the following quantitative analysis was to define different categories of franchisee based on their perception of autonomy and performance. We created a questionnaire, send them, received and treated 226 of them, completed by franchisees. We thus conducted one multiple correspondence analysis (MCA) on the autonomy perception indicators and another MCA on the performance perception indicators. As a results, we found some critical success factors for franchising: the sense of attachment of the franchisee to the network, of promoting and recognizing franchising initiatives in store management activities and in dealing with local competition. Also critical were : improving franchisee capacity of training their staff and giving autonomy to franchisee in these training activities. Finally we found a positive relationship between autonomy and performance, the equilibrium between autonomy and competence being an essential condition of performance.