This thesis examines how power is constituted in hybrid polycentric systems of security governance. In particular, the thesis explores how legitimacy - as one form of power - is configured in Improvement Districts in South Africa, with a specific focus on three ways by which it is gained: through promoting public participation in decision-making; through transparent and accountable policing nodes; and through the delivery of effective security for the public good. Polycentric systems of security governance are usually composed of a number of policing or security nodes that are independent of each other, but take account of each other in relationships of co-operation or conflict and where no single node dominates all the rest. In other words, some or all of these nodes, may co-ordinate around specific security problems or events in a sustained manner. The functioning of polycentric security governance was explored in Improvement Districts in Cape Town and Johannesburg, as they are an exemplar of polycentricity in the way that they operate. Qualitative field research was employed using a nodal analytical framework and a collective case study approach. In-depth interviewing, participant and direct observation as well as documentary analysis were the primary research methods employed. The findings of the research reveal that polycentricity impacts on legitimacy in a number of ways. Legitimacy may originate from multiple sources and state and non-state policing nodes within polycentric security governance systems may undermine, enhance and/or co-produce democratic participation, accountability and security for the public interest. There are a number of factors or conditions that shape whether polycentric systems of governance are legitimate and how they derive this legitimacy. The main finding of the thesis is that for a polycentric system to be aligned to the public interest, it needs to be motivated by public, peer and political expectations, amongst other things. The findings of the thesis both challenge the normative tendency to associate democratic legitimacy with the state and contribute to the pressing question of how to theoretically account for the empirical reality of polycentric security governance systems.