ABSTRACT The fluvial flood disaster (FFD) poses a great threat to human civilisation. The high water level can cause overtopping, piping, and even breach of the levees, which, in turn, further results in urban flood. These secondary disasters form a disaster chain with multipaths that amplify the risk of the fluvial flood. This work develops a stochastic multi-path network method to quantify the risk contagion for the FFD chain. First, the multi-path network structure for the chain was identified, characterised by reflecting the uncertainties of levee failure modes. Second, a model for the evolutionary probabilities of the chain was proposed by combining the directed network technique with the network structure. Finally, the systematic risk of the chain was quantified by considering the consequence of the disasters. A risk contagion factor was proposed to quantify the attenuation or amplification effect of the risk during the evolution of the chain. A case study in Huizhou, China, shows that the FFD chain is characterised by “small probability” but “extremely high risk”. The most likely path is overtopping-urban flooding, indicating the priority of the area is to heighten the levees. This work paves the way for a more entrepreneurial future on disaster prevention and mitigation.