According to the resource curse, oil revenues constitute a lucrative prize, motivating rebel groups to fight for control of resource-rich territory. Yet rebels rarely attack sites with the most oil, such as oil fields and terminals. To explain this finding, we develop a new model with elements of crisis-bargaining and Blotto games and use geo-spatial data on the location of oil infrastructure and armed conflict events to assess its implications. First, we show that rebels focus their attacks on pipelines and that more critical infrastructure has no discernible effect on conflict. Second, to prevent the government from thwarting their attacks, we find that rebels strategically randomize where they strike. Third, we find that increased oil prices have countervailing effects: as the black-market price increases, so too do the returns to oil theft; but as the export price increases, violence abates because, we argue, governments are more eager to ‘bargain away’ conflict. Overall, groups sabotage pipelines because they expect these sites to be vulnerable and because such disruption can compel governments to address their demands.