What conditions lead to confidence among civil war combatants that they will not face accountability for perpetrating sexual violence? This study investigates the causes of impunity for sexual violence among armed actors. It develops a theoretical framework which identifies three explanations for armed group impunity for sexual violence, namely (1) flawed prohibitions inside an armed group; (2) negligent enforcement by its authorities; and (3) pardons in the form of amnesties during the peace process. Adopting a two-pronged approach, the study first explores the associations between amnesties arising from concluding peace agreements and post-settlement levels of sexual violence in Burundi, the Democratic Republic of Congo, Liberia, Mozambique, Sierra Leone and South Africa. A small-scale, events-based dataset of sexual violence by governments and rebel groups in the first three years after war was constructed. The second and main part of the study is a comparison between two rebel groups in Burundi’s civil war (1994-2008), CNDD-FDD (National Council for the Defence of Democracy-Forces for the Defence of Democracy) and Palipehutu-FNL (Palipehutu-Forces for National Liberation) and their practices of prohibition and punishment of wartime sexual violence, taking into account also the possible influence of amnesties. Based on original data from focus groups of ex-combatants from these rebel organisations, it is found that flawed prohibitions and negligent authorities are the main explanations for armed group impunity. The findings do not support amnesties as a cause of armed group impunity for sexual violence. Moreover, additional findings suggest that accountability for sexual violence is triggered by dependency on civilian support, while impunity is facilitated by an armed group’s ability to secure recruits, material and other resources without the help of local communities.