This dissertation is a study of local-level variation in electoral violence. Kenya has experienced recurring and large-scale violence, yet there are striking differences in the sites and scale of violence within each electoral period. Why does violence escalate in some contexts of Kenya, and not others? Elite-centered approaches to election-violence have identified important cross-national and temporal patterns, yet overlook how the electoral incentives of elites interact with the interests and motives of ordinary citizens. As a result, most studies do not identify the local dynamics that underlie the process and organization of electoral violence---both where it escalates and where it does not. The dissertation demonstrates that variation in land rights between groups shapes the process and patterns of electoral violence across Kenya. By focusing on the politics of land security, I identify the mechanisms that link the motives and methods of elites with the interests of ordinary citizens. I argue that the escalation of violence is a process that relies jointly on three main factors: 1) land rights inequality between "insiders" and "outsiders," 2) contentious land narratives between these two groups, and 3) the ability of political elites to use these narratives as a tool to organize election violence. Electoral violence is much more likely when elites can draw on land narratives to convince followers that elections present a credible threat to land rights or an opportunity to re-claim land. The evidence for this dissertation draws from 15 months of multi-method research that I conducted in Kenya. The first stage of the research is a micro-comparative case study across government settlement farms and land buying cooperatives (LBCs) in the Rift Valley and Coast regions. In total, I interviewed 230 ordinary Kenyans and conducted twenty focus groups. The second stage is a household-level survey that I designed and administered to 750 Kenyans. The survey measures individual-level land security, belief in contentious land narratives, and election violence, while accounting for other factors.