The impact of a sudden, mass influx of forced migrants on the hosting economy is understudied and not well understood. Using a household panel dataset for the Kagera region in Tanzania, we test the impact of the mass refugee presence on the welfare of the local population. Our identification strategy exploits variations in the timing of arrival as well as the spatial distribution of Rwandese and Burundi refugees. We find a positive and aggregate effect, but households are impacted differently depending on their main initial occupation.