Nowadays when confronting natural disasters, emergency evacuations may occur. Household evacuation decision modelling is frequently used to estimate the potential number of non-self-evacuated persons to care for when planning resources for an effective disaster management preparedness. Indeed, predicting households behaviour in the case of the different factors (evacuation order issued by the authorities, critical infrastructures failures like power outage or sewage treatment overflow, social influence, etc.) that could influence their willingness to leave their home in the context of natural disasters is meaningful for authorities and operational entities to make specific evacuation plans. Since households evacuation in the face of flood induced disasters could be motivated by several factors, a departure response curve that states the percentage of departures at a given interval of time is typically required for each factor separately. However, the existing behavioural response curves are mostly linked to an evacuation order and it does not exist in the literature studies which report the availability of response curves for critical infrastructures failures. This communication aims to present an empirical study conducted for investigating evacuees’ behaviour under power supply failure conditions due to flood disaster. This study is based on survey data collected in the Ile-de-France region following the recent historical floods events of May-June 2016 and January-February 2018 in the Ile-de-France region (France). The findings of this study may be useful for modelling self-evacuation decision-making that considers power outage as a triggering factor. This study also provides insight into the need of providing modellers with a specific departure response curve for each of the factors that could influence evacuation decision-making through future works.