This thesis looks at the practices that classify ‘fragile states’ and their impacts. I analyse not only direct practices of ranking and categorisation but many indirect practices that make of ‘state fragility’ a measurable and manageable political truth. This research looks at the relations between the World Bank, OECD and the g7+ group of self-labelled ‘fragile states’ to understand how this measurable and manageable political truth is constructed, what power this construction involves and what its implications are. ‘State fragility’ has been largely quantified in systems used to compare state performance. Drawing from Bourdieu’s sociology and Hacking’s philosophy, I suggest these practices of quantification and classification are not a mere technicality; they answer to deep-seated successes of statistical reasoning, and their political entrenchment in policy-making has particular impacts in the ‘fragile states’ agenda. I suggest quantifying and classifying practices constitute a style of thinking and doing that carries symbolic power, a subtle but no less important form of power. Symbolic power is diffuse, and through the subtlety of practical sense it wins the complicity of those who seem least favoured by it – as crucially and richly exemplified by the self-labelling and self-measurement of the g7+ in terms of ‘state fragility’. However, in the diffuse and hardly traceable form of statistics, it also carries the elements that allow its own weakening and change. While direct power hardly leaves arms unattended, quantification by necessity travels with its tools and practices, hence, making it more possible to catch hold of some of these elements. I argue that by ignoring the practical sense in quantification and classification and by taking for granted what power is in these dynamics, how it is exerted and by whom, many critics paradoxically side-line the very subtle but important possibilities for weakening and change this symbolic power carries within.