This dissertation examines the philosophical foundations of social choice theory. Social choice theory is the area of normative economics which is concerned with the aggregation of individual preferences. The aim of this work is to investigate the philosophical assumptions of social choice theory in order to understand to what extent it can contribute to a theory of justice based on capabilities. Therefore, the dissertation is build up on Amartya Sen’s idea of a “comparative approach” of justice, as opposed to the Rawlsian “transcendental approach”. It is an attempt to precise which understanding of social choice theory is required to specify the capability approach, especially the evaluation and the indexing of capabilities. In this dissertation, we argue that the apparent tension between preference aggregation and capability approach is due to a narrow interpretation of social choice theory’s conceptual framework. Capability approach is generally conceived as non-compatible with the preferentalism of social choice theory: after all, capabilities are seen as a response to the recurring problem of adaptive preferences. This dissertation thus consists in widening the scope of interpretations of social choice theory framework. This research deals mainly with the informational basis of social choice theory.This dissertation is in three parts. The first part tackles the following problem: are preferences determined by an individual source that can be thought independently of its social and economic position? To answer these questions, three kinds of informational basis in social choice theory and normative economics are investigated: cardinal utilities, ordinal preferences and capabilities.The second part aims at defining what preferences do describe in this context. Firstly, the nature of preference itself is examined: can it be assimilated to choice? Or is it a mere evaluation? A desire? A mental state? This analysis points out the comparative structure of preferences. Secondly, the various ethical criteria of preference are investigated: hedonistic pleasure, desire satisfaction and objective well-being. We argue that preferences are better conceived as comparative evaluation and require actually excellent cognitive conditions to be truly the agent’s own real preferences.The third part goes back to capability approach. The argument relies on the previous results to build up a first account of a democratic non-ideal theory of justice based on capabilities. In this part, we show that a preference-independent capability evaluation turns out to dismiss the importance of freedom and agency in capability approach. Then we argue that capabilities and functionings as an object for preferences do provide a first filter against adaptive preferences.