This paper tackles the problem of the demarcation of value judgments in economic expertise. Is it possible to disentangle values from facts, or neutral scientific assertions from value-laden judgments, in the context of economic expertise ? If not, why not ? And if it is, under what conditions ? First, drawing on concepts from analytic philosophy, the paper highlights the interdependencies between descriptive, evaluative, and prescriptive judgments. Second, drawing notably on social studies of science, the paper proposes a definition of 'expertise', and translates this into a list of successive stages wherein these different types of judgments are involved. A backward analysis of these stages is provided in order to identify where values stand, and who holds them. Third, reconsidering the positions of neutrality in economics (Mongin 2006), the paper defends the 'weak non-neutrality' view in the context of expertise, and concludes with reflections on what could be done to address the problem of democratic legitimacy raised by the difficulty of demarcation.