This paper presents the application of a graph-based methodology for the assessment of flood impacts in an urban context. In this methodology, exposed elements are organized as nodes on a graph, which is used to propagate impacts from directly affected nodes to other nodes across graph links. Compared to traditional approaches, the main advantage of the adopted methodology lies in the possibility of identifying and understanding indirect impacts and cascading effects. The application case concerns floods numerically reconstructed in Mexico City in response to rainfall events of increasing return periods. The hazard reconstruction was carried out by using a simplified hydrological/hydraulic model of the urban drainage system, implemented in EPASWMM, the Storm Water Management Model developed by the United States Environmental Protection Agency. The paper shows how the impacts are propagated along different orders of the impact chain for each return period and compares the risk curves between direct and indirect impact. It also highlights the extent to which the reduction in demand of services from consumers and the loss of services from suppliers are respectively contributing to the final indirect impacts. Finally, it illustrates how different impact mitigation measures can be formulated based on systemic information provided by the analysis of graph properties and taking into account indirect impacts.