In this paper we examine the impact of precipitation variations on the probability of conflict in Ethiopia using subnational data at 0.5×0.5 decimal degrees resolution for the period 1997 to 2013. We find that lower precipitation levels, after accounting for the long-term average, are associated with higher probability of conflict. Our results are robust to alternative model specifications. The impact of precipitation on conflict remains significant for intra-state conflict but loses significance for non- state conflict. Moreover, using a two-stage estimation method we find evidence for the hypothesis that precipitation affects conflict through affecting total production levels.