Flood-risk assessments are an objective and quantitative basis for implementing harmonized flood mitigation policies at the basin scale. However, the generated results are subject to different sources of uncertainty arising from underlying assumptions, data availability and the random nature of the phenomenon. These sources of uncertainty are likely to bias conclusions because they are irregularly distributed in space. Therefore, this paper addresses the question of the influence of local features on the expected annual damage in different municipalities. Based on results generated in the frame of a transnational flood-risk-assessment project for the river Meuse (Western Europe) taking climate change into account, the paper presents an analysis of the relative contributions of different sources of uncertainty within one single administrative region (the Walloon region in Belgium, i.e. a river reach of approximately 150 km). The main sources of uncertainty are not only found to vary both from one municipality to the other and in time, but also to induce opposite effects on the computed damage. Nevertheless, practical conclusions for policy-makers can still be drawn.