Since Collier and Hoeffler (1998, 2004), it has been supported that inequality, measured at national level, does not affect the risk of conflict. Such a result has been much debated in the literature. Based on a revisited theoretical framework, the purpose of the paper is to explore the role of inequality in localized conflicts. We argue that previous findings might be biased by the myopic nature of cross-country analysis. Consistently with the model, Probit estimations indicate that income inequality measured at municipal level was significant in motivating people to support the rebellion in South Mexico. At this geographical level, we also find an increase in income per capita could exacerbate the risk of conflict in a situation where the rebel leader would have greater incentives to loot the local production compared to the opportunity cost associated with fighting for the worker.