This dissertation draws on a total of twenty-seven months of longitudinal ethnographic research conducted over a six-year period investigating the impacts of a proximity policing initiative known as "pacification." Implemented in Rio de Janeiro beginning in 2008, this initiative aimed to regain the state monopoly of violence in favelas dominated by armed drug dealers while also reducing police violence. In early years, groups such as the World Bank viewed pacification as a successful policing program worthy of emulation in criminal-dominated urban peripheries throughout the global South. However, by 2016 it was clear that it had failed, as even the most successful "model communities" experienced an unparalleled spike in violence. What explains this outbreak of violence, especially in the favelas where the program was initially well-received and appeared to be working? What went wrong?Due to limited and problematic police data, biased media coverage, and a rapidly changing empirical reality, it is very difficult to know what is actually going on in these communities from the outside. In 2012, I began ethnographic research in two neighboring "model communities" to try to understand what was working there that set them apart from other "pacified" favelas. As I returned in the summers of 2013, 2014, and 2015, I observed the rapidly shifting community relations as residents, police, and drug dealers alike prepared for what they perceived as the approaching expiration date of pacification, set by the 2016 Summer Olympic Games. By the time I arrived for a full year of fieldwork in 2016, residents were saying that pacification is not going to end, it has already ended. Rather than the single violent post-Olympic invasion that residents had imagined, 2016 was characterized by a series of invasions, attempted invasions, targeted assassinations, and killings by police as drug dealers from both factions called on a seemingly endless supply of outside reinforcements from allied favelas. I elaborate a theory of intersecting temporalities to make sense of these community-level changes framed by what I call mega-event time. While state time is largely cyclical, mega-event time consists of a build-up period to a single major event. Increased investment, accelerated urban restructuring, and heightened international media attention characterize this period. The event itself is marked by all-out international performativity. After the event has passed, the performance ends – unless there is another event immediately on the horizon. During the period of my research, Rio hosted a series of back-to-back events, forming a continuous build-up to the final and biggest event in the series – the 2016 Summer Olympic Games. I argue that the spike in violence in 2016 and after was the direct result of a public security initiative implemented on mega-event time that had, as residents say, a prazo de validade [expiration date]. My research deepens interdisciplinary discussions of uneven development and state presence/absence in marginalized communities. It suggests that rather than talking about state absence in these areas, we need to examine (and empirically specify) convoluted forms of state presence that are deeply intertwined with the functioning of criminal groups. My dissertation also places public security reform in global context. It contributes to the growing concern about the effects of mega-events on host cities, especially in the global South. Temporary public security interventions can have disastrous consequences for marginalized communities even when carried out under the guise of promoting human rights.