A social policy reform aiming at reducing poverty by drafting a new minimum income scheme (MIS) can be a very sensitive issue. It causes reluctance and controversies in policy arenas as in public opinion. Studying the political usages of a specific evaluation is a good way to better understand its institutionalization's process. Hence, it's interesting to analyze how and why the policymakers call on evaluation in the policymaking process. It brings to seize the role played by evaluation in the production, the legitimization and even the reframing of a social policy. The "Revenu de solidarité active"(RSA) is a French MIS designed for the unemployed and poor workers. I propose to analyze two RSA's evaluations : one has been conducted on the RSA experimentation (2008) and the second one was targeting its generalization (2011). I conducted a research on both evaluations with some fifty interviews of political and administrative actors, and a participant observation. This paper is based on a doctoral research about the institutionalization of social policy's evaluation, its stakes and usages in the family and social branch of the French social security. First, it analyzes the composition of the evaluation comity and its interinstitutional perspective, involving different actors, and implying specific perceptions and practices. Second, it shows how the first evaluation has been used by political actors to legitimate this MIS through a depoliticization of the debates, despite the weak evidence of the RSA's efficiency. Then it study how the second evaluation partially reconsiders the RSA's efficiency revealing an important non-take-up phenomenon that still does not lead to question this MIS relevance. More broadly, this paper tries to understand how the political actors use the evaluation practices in the social policy area. Thus it aims at studying the influence of a policy sector through its specific set of actors, controversies and administrative culture upon the evaluation practices and usages.< Réduire