Vigilante violence is generally understood as an alternative means of controlling crime and providing security where the state does not. It has been found in nearly all modern societies at one point or another. Currently, in South Africa, vigilantism is common, accounting for roughly 5% of daily homicides. Despite its ubiquity, vigilante violence has largely been ignored by scholars, and in South Africa, vigilante violence tends to be dismissed as “mob violence.” This dissertation draws on extensive fieldwork, multiple qualitative and quantitative data sources, and different theoretical and methodological approaches, to provide a comprehensive analysis of vigilante violence in Gauteng, South Africa. The first paper address critical theoretical issues surrounding the role of weak and failed states in fostering vigilantism. In this analysis, I use large-scale quantitative data from the Gauteng City-Region Observatory 2013 Quality of Life Survey and an independently compiled database of newspaper articles detailing incidents of vigilantism in Gauteng. I employ measures of perceptions of government performance and the provision of state security to test the relationship between perceived state legitimacy and vigilante violence. I find that negative perceptions of government performance are actually associated with decreases in vigilante violence, while negative perceptions of state security are associated with increases. The second paper utilizes the same data sources and uses the well-establish social disorganization ad neighborhood effects literature to examine the relationship between neighborhood cohesion, collective efficacy, and vigilante violence. I find that, in contrast to existing research, higher levels of neighborhood cohesion and collective efficacy actually result in more incidents of vigilante violence. The third paper expands upon the micro-sociological perspective of violence developed by Collins (2008), “forward panic,” the process whereby the tension and fear marking most potentially violent situations is suddenly released, bringing about extraordinary acts of violence. Analysis of in-depth interviews shows that episodes of vigilante violence in townships are often clearly episodes of forward panic. Although the concept of forward panic focuses on individuals, I argue that if the pre-conditions that foster forward panics in individuals are structural, there is the potential for forward panic in entire groups or parts of communities.