This paper examines how PISA surveys have been received and employed in French-speaking Belgium. It involves studying the application of highly specific instrument to a context whose particular characteristics have been forged over time. After pointing out the historically specific attributes of consociational politics in Belgium, and the traditionally modest place of knowledge in the making of public policy (Mangez, 2009, 2010), we will focus on the PISA instrument. What kind of instrument is it? Does it resemble a mechanism of the Evaluative State, that is, is it an instrument useful in decision-making? Alternatively, does it perhaps constitute an instrument required for surveillance in the way Rosanvallon (2006) uses the term? Perhaps should we view it as an instrument facilitating soft regulation? Unlike the official position, which views PISA as an instrument supporting decision-making, our findings indicate that the instrument — at least in the context of French -speaking Belgium — tends instead to function as a monitoring instrument that increases the pressure of the quasi -market on actors (public and private) in the educational sector. According to this analysis, the State should not be viewed solely as an evaluative State; it must also be evaluated, ranked and compared to other purveyors of educational services, both Belgian and foreign. The analysis prompts us to advocate a new concept, that of « third-party evaluator » (Rosanvallon, 2006), which could enrich and refine analysis of the role of knowledge in public policy.