The impact of a mass influx of refugees on the hosting economy is understudied and not well understood. Based on hypotheses formulated during two months of fieldwork and using a theoretical framework, we test the impact of the refugee presence on the welfare of the population of Kagera, Tanzania. Our identification strategy exploits both time and spatial variations in the way households traced between 1991 and 2004 have been affected by the refugee inflows originating from Burundi (1993) and Rwanda (1994). While the aggregate effect is positive, local hosts, depending on their occupation, can be winners or losers.