This paper investigates the relationship between increasing inequality and recruitment in a civil conflict. Starting from the puzzling observation that the Nepalese conflict escalated after a period of substantial growth and poverty reduction, it hypothesizes that increasing differences in welfare between groups can help explain recruitment by the Maoists. The hypothesis is tested with data from two national-representative household surveys, matched with district-level information regarding mass abductions by the Maoists from newspaper articles. The identification strategy relies on the fact that the months following finalization of the second round of data collection were characterized by a geographical escalation of the conflict. The paper first shows that gains from growth between 1995 and 2003 were much smaller for the (near) landless than for the landed; it then shows that recruiting through abduction of young people was more intensive in districts where inequality between the landed and the landless had previously increased.