International interventions in mass killing episodes often fail to adequately protect civilians. The UNMISS peacekeeping operation in South Sudan is a case in point. Bureaucratic, political, and financial constraints consistently inhibit the deployment of well-staffed interventions, and often prohibit them outright. Therefore, many civilians must survive without external assistance, but we know surprisingly little about how this occurs. No scholar has produced a work combining empirical examples of civilian self-protection with a theory of the mechanisms that allow it to function during mass killing or even conflict more broadly. The lack of focus on civilian self-protection is symptomatic of a larger issue of how scholars envision violent conflict. Conflict’s inherent complexity makes it analytically less difficult to envision armed actors as the main influences on the course of conflict, rather than civilians, who are more numerous and whose motivations are less explicit (though it should be noted the civilian/combatant dichotomy is fuzzy). Without an understanding of civilian action during conflict, it becomes difficult to imagine how civilians protect themselves. However, the last 10 years have seen a shift, albeit incomplete, toward a research agenda focused on civilians as agents of conflict and conflict resolution. Stathis Kalyvas’s The Logic of Violence in Civil War argues for a new reconceptualization of civil wars as driven by civilian action. Building on Kaylvas’s work, Oliver Kaplan’s scholarship on civilian autonomy in Colombia highlights the ways in which civilians protect themselves through small-scale organization. Additionally, Casey Barrs has sought to understand the specific dynamics of civilian self-protection and how outsiders can aid this phenomenon. It is principally from these two bodies of work that my undergraduate thesis, “Filling the Gap: Nonviolent strategies for civilian protection during mass atrocities” draws from. My thesis provides a theoretical and empirical grounding for nonviolent civilian self-protection. I make four central claims about the nature of civilian self-protection. First, most self-protective actions are nonviolent. While civilians do sometimes protect themselves violently, most do not have the capacity to do so effectively. Violent tactics are also combined with a myriad of nonviolent tactics. Second, civilians survive with the aid of small-scale social networks. Every existing social structure takes on a protective capacity as civilians seek to negotiate with combatants, gather resources, and/or facilitate flight. Third, local knowledge is key. During the LRA insurgency in Uganda, Baines and Paddon note that civilians developed sophisticated early warning networks and relocation plans using their superior knowledge of local human and physical geography. Finally, civilians transmit norms of behavior to soldiers. Combatants are often sensitive to civilian opinion for emotional, moral, and strategic reasons, and civilian appeals can facilitate rapid changes in combatant behavior. Many scholars have dismissed nonviolence as a way for civilians to protect themselves from mass killing; however, this viewpoint ignores the process through which violence develops. Mass killing occurs through the escalation of violence, but it is very rare for an episode to reach the level at which combatants have the ability and desire to kill all non-allied civilians in their area of operation, leaving space for civilian self-protection. Civilian agency, however, does decrease as conflicts intensify; violence rather than the provision of benefits becomes the central avenue for exerting power. Other scholars, such as Semelin and Cormier et al., argue for nonviolent civil resistance in response to genocide and mass killing. The problem here is that during mass killing, the existing space for civilian nonviolent action is not large enough to facilitate large-scale resistance. Combatants will not tolerate explicit oppositional civilian organization, and therefore attempts to do so will likely be suicidal. The dynamics of civilian self-protection is not dissimilar to civil resistance, but the increased potential for repressive violence in the context of mass killing limits civilians to small-scale organization and apolitical goals. In a cross-national study, Local to Global Protection found that self-protective ability is not evenly distributed across communities experiencing conflict. Therefore there is potential for outsiders to improve protective capacities through establishing effective localized early warning systems and rapidly-deployed survival aid. However, for these practices to be widely adopted and accepted, much more research on civilian self-protection is needed. Moving past an external assistance-focused research agenda and top-down approaches to civilian protection is crucial for scholars and practitioners hoping to lessen the terrible toll of mass violence. Danny Hirschel-Burns is a 2014 graduate of Swarthmore College and former National Policy Coordinator for STAND.