The impact of hosting refugees on child labor in host countries is unclear. This paper estimates both the short and the long term consequences of hosting refugees fleeing from the genocides of Rwanda and Burundi in the Kagera region of Tanzania between 1991 and 2004. The study uses longitudinal data from the Kagera Health and Development Survey. Using the exogenous nature of refugee settlement in Kagera due to geographic and logistical reasons, we find the causal impact of hosting refugees on child labor and children's schooling outcomes. The results suggest that the impact of hosting refugees on children living in Kagera decreases child labor in the short run (between 1991 and 1994), but increases it in the longer run (1991-2004). The results are heterogeneous across gender and age. The study aims at understanding the mechanisms behind the variation in child labor outcome due to the forced migration shock exploring various channels.