This dissertation develops a theory on the strategic interaction between civilians and combatants in war zones. It builds on the emerging field of research on noncombatant self-protection mechanisms in civil wars and addresses two questions: first, once civilians have decided to organize a resistance campaign, why do they choose armed or unarmed methods of struggle? Second, how do armed groups respond to this challenge? Regarding resistance strategies, I argue norms against the use of violence determine the content of a campaign when a community depends on an external ally to organize the mobilization and such an actor has strong preferences for peaceful activism. Strategic factors (e.g., rebels’ reputation) are determinant when norm entrepreneurs are absent or fail to influence the community. Concerning groups’ reaction to a resistance campaign, the dissertation conducts one of the first comparisons of the effectiveness of these strategies. To do so, it advances a typology that distinguishes between moderate (i.e., protests) and radical strategies (either unarmed or armed self-protection) and proposes two game-theoretic models of the civilian-combatant interaction in war zones. These models predict that rebels are more likely to repress radical rather than moderate strategies. The dissertation tests this theory with statistical methods and a novel dataset on resistance campaigns in Colombia (1985-2005). It explores the Catholic Church's promotion of peaceful activism against the war in this country and tests the theory on the civilian choice of strategy with multilevel multinomial models. This empirical method yields robust evidence to the theory: when encountering a rebel group with a negative reputation, civilians are more likely to organize violent self-protection rather than peaceful activism. In contrast, when there is a powerful third party with the leverage and resolution to promote nonviolent action, civilians are more willing to undertake a peaceful mobilization rather than create a militia. The empirical expectations concerning the warring parties’ reaction to civilian dissent are examined with methods for causal inference with panel data. The dissertation corroborates that protests can compel insurgents to make concessions to the population. In contrast, rebels tend to harshly retaliate against communities that escalate a campaign with violent tactics.