Social scientists have studied why rebels deliberately kill civilians. Explanations of rebel group propensities to target civilians include reference to economic factors, organisational dynamics, and military power asymmetries. What binds this diverse scholarship together is the premise that armed non-state actors use violence against non-military targets strategically to gain and sustain power. Instead of analysing why and when targeting civilians during wartime can benefit rebels, this PhD seeks to build a political case for restraint in warfare. The aim is to understand the conditions under which restraining violence against noncombatants makes political sense for armed non-state actors. According to Gene Sharp, a leading theorist of nonviolent resistance, the power of rulers in any political system depends on the consent of the ruled. By withdrawing that consent, the ruled can control and ultimately destroy their rulers’ power. Sharp’s work, followed by subsequent generations of resistance scholars, was essentially state-centred. The main objective was to understand how people can limit state power and eventually oust repressive regimes. More recently, scholars have invoked the concept of resistance to examine unarmed action by civilian populations against armed non-state groups. Growing interest in civilian resistance during armed conflict is premised on the assumption that local populations are often better placed than external actors, such as international humanitarian organisations, to protect themselves against rebel violence and abuse. Whereas externally-led efforts to protect civilians from violence draw on legal obligations enshrined in international humanitarian and human rights law, civilian resistance against rebels is essentially a political process. If rebel rulers, just like states, depend on support from populations they claim to control, they may have a political incentive to accommodate civilian demands made in a nonviolent manner. Sometimes labelled the ‘new frontier’ in resistance studies, nonviolent action against rebels is an emerging research field that remains in its infancy. Some significant conceptual contributions have reduced overlap between nonviolent resistance and related concepts, such as civilian noncooperation with armed actors. However, much existing work on nonviolent action against rebels is either descriptive or focused on the question of when civilian resistance may emerge. Little systematic research has yet examined the conditions that enable unarmed civilians to wring concessions from armed groups. This PhD addresses this gap in three articles. By scrutinising the effectiveness of civilian resistance, I provide a necessary corrective to the dominant view in the literature, which, as my research suggests, is overly optimistic regarding prospects of wringing substantial concessions from armed groups via nonviolent action. My first article, Between ‘Flight’ and ‘Fight’: When Does Civilian Resistance Against Rebels Work?’, is a conceptual work that reviews the growing literature on civilian resistance against rebel groups. Drawn from an extensive review of academic articles, book chapters and policy documents, I order examples of civilian resistance into distinct categories of unarmed action: deception, dissent, and defiance. This classification provides the conceptual framework to tackle the main research question: Does civilian resistance against rebels work to protect unarmed populations from violence and harm? After analysing successful and unsuccessful examples of civilian resistance documented in 21 different country contexts, I identify common factors that have shaped rebel responses to civilian resistance. The aim is to contribute to theory-building by developing hypotheses, which I have subsequently refined and tested in my empirical research. The article concludes by discussing policy implications, focussing on normative challenges for external actors eager to support civilian resisters. My literature review reveals a geographic cluster: of the 64 cases of civilian resistance documented in the literature, almost one-third concern one country: Colombia. With 25 cases, Colombia is by far the most researched country context, followed by Syria (8 cases). Within literature on Colombia, most scholars focus on civilian resistance in subnational areas where rebels competed for power and control with other armed non-state actors. Moreover, half of the works on Colombia (17 of 25) explore areas with proportionately larger indigenous populations. The particular conditions that have enabled indigenous people to resist relatively weak rebels are not necessarily present or replicable elsewhere. The observed bias in existing research towards places that provided fertile ground for civilian resistance constituted the starting point for my second PhD article, ‘Rebels Do Not Take Kindly to Criticism: The Strategic Failure of Local Resistance Against Colombia’s FARC’. Using Colombia’s Caquetá region as a ‘hard case’, the article asks what factors influence the success or failure of civilian resistance against rebel groups under unfavourable conditions. For several reasons, the Caquetá region poses a difficult test for the prevailing idea that armed groups often respond positively to unarmed pressure. Situated in the country’s south, Caquetá has no sizeable or politically organised indigenous community. Moreover, Caquetá is among the most emblematic historic strongholds of the Fuerzas Armadas Revolucionarias de Colombia (FARC). Founded in 1964 as an umbrella organisation of various local peasant defence forces, the FARC was for long Latin America’s oldest and most powerful guerrilla organisation. The FARC’s history is intrinsically tied to Caquetá, which served both as a strategic rear guard and, more importantly, as a showcase for the group’s ability to govern civilian populations. Public displays of civilian discontent under rebel rule were at cross purposes with the FARC’s political ambitions in this part of Colombia. My empirical investigations, which relied on ethnographic methods and archival research, yielded more successful examples of civilian resistance in Caquetá than I expected. Even under unfavourable conditions, civilians in Caquetá managed to wring concessions from the FARC via nonviolent resistance: travel restriction exemptions for teachers and health workers; access permits for humanitarian agencies; and the removal of abusive local commanders. While these achievements are remarkable, the underlying civilian demands do not qualify as maximalist because they did not clash with the FARC’s strategic interests. As soon as civilians openly defied the FARC by disregarding with the group’s political directives, the customary response was repression. I demonstrate that this was the case when civilians chose to run for political office and campaign in Caquetá despite a FARC-declared electoral boycott. One in four civilians killed by the FARC between 1990 and 2010 was a local politician who defied electoral boycotts or refused to step down despite a FARC order. In addition to defying electoral boycotts, civilians also resisted FARC effort to influence elections to its advantage in less violent ways by promoting supportive political candidates. Civilians interviewed in Caquetá related several incidents when the FARC called meetings in advance of local elections to ‘preselect’ candidates. Although civilians tended to endorse FARC-proposed candidates in these public meetings, they behaved differently at the ballot box, casting their votes for candidates who opposed the FARC. This was a clear and particularly embarrassing rejection of the FARC’s authority. Such hidden forms of electoral resistance were only a foretaste of what awaited the FARC after it laid down its arms. As part of a peace agreement signed in 2016, the FARC transformed into a political party, which initially retained the guerrilla group’s wartime acronym. During the first post-agreement election in early 2018, the FARC party earned a meagre 0.3% of the vote, relegating the group to the side-lines, unable to challenge the Colombian government against which it had fought for over half a century. Journalists and academics alike were quick to attribute the FARC party’s dismal electoral début to the guerrilla group’s wartime abuse of civilians. Yet, if excessive violence against civilians reduced the FARC’s electoral support, one would expect the negative correlation between violence and vote shares to hold at a subnational level. Available micro-level data on FARC violence and pro-FARC votes in the 2018 election is more ambiguous. Whilst the FARC scored badly overall in 2018, the party did comparably well in areas that had seen high levels of FARC violence against civilians, garnering up to 18% of votes. Why did the ex-rebels score best in subnational areas that saw high levels of FARC wartime violence? This the main puzzle of my third PhD article, a co-authored and quantitative piece: ‘Fighting for votes: Wartime violence and electoral support for Colombia’s FARC rebels’. Simply put, there are two possible solutions to the puzzle. The first is that Colombia’s 2018 election disproves the intuitively plausible hypothesis that civilians punish rebels for excessive wartime violence at the ballot box. If this were the case, wartime violence would have had little or no effect on electoral behaviour following the FARC’s transformation into a political party. But our multiple regression analysis of subnational variation in pro-FARC vote shares point to a second explanation: those who voted for ex-rebels, regardless of or despite violence, were essentially ex-fighters and members of the FARC’s clandestine support networks. Support for the FARC party in 2018, however, exceeded 5% in only 14 municipalities, which corresponds to roughly 1% of Colombia’s 1,122 municipalities. In the remaining 99%, the FARC suffered crushing defeat. The guerrilla group’s soaring record of violence against civilians since the 1990s, as we argue, remains a plausible explanation for why the FARC party lost its recent fight at the ballot box. The FARC’s growing reliance on repugnant tactics, particularly civilian kidnappings, reduced the ex-rebels’ ability to mobilise electoral support in 2018 beyond its immediate support base. This should serve as a reminder to contemporary rebel groups that excessive wartime violence against civilians is likely to backfire in peacetime. If wartime violence reduces rebel groups’ chances of post-war political representation, political arguments may help to persuade conflict parties to spare civilian lives. Such a nonlegal discourse could complement law-based advocacy efforts, particularly regarding rebel groups, which are by definition not party to international treaties such as the 1949 Geneva Conventions. To make a strong political case for restraint in war, however, we need to demonstrate more clearly that violence against non-military targets generally translates into reduced support at the ballot box. An obvious limitation of the conclusions I draw, and of other case-based studies, is their limited generalisability. One way to increase external validity is to assess the effects of violence on electoral support via large-n studies or cross-country comparisons. To establish whether political success is indeed linked to restraint, beyond individual cases, we must develop conceptual standards and measurements that enable comparison of degrees to which armed groups target civilians across cases.