There is ample evidence to show that choice behavior often deviates from the classical principle of maximization. This evidence raises at least two important questions: (i) how severe the deviation is and (ii) which method is the best for extracting relevant information from the choices of the individual for the purposes of welfare analysis. In this paper we address these two questions by proposing a set of foundational conditions on which to build a proper measure of the rationality of individuals, and enable individual welfare analysis of potentially inconsistent subjects, all based on standard revealed preference data. In our first result, we show that there is a unique measure of rationality that satisfies all of the proposed axioms: the weighted-loss indices. In the second part of the paper, we study some relevant properties of weighted-loss indices.