Since the 90’s, vulnerability indices have emerged as a leading tool to quantify and map the dimensions of hazards vulnerability. However, this quantitative approach is criticized for giving only a rather general overview of the situation, without rendering the complexity of the phenomena involved or the local context, because they are not always adapted to the changes in scale and seem dependent on the initial choice of variables. Moreover, the typical rationale for decisions regarding variable selection, analysis scale or method is based on simplicity or choices made in previous studies, and in many cases, no justification is provided at all. Furthermore, these quantitative approaches are almost never confronted to feedback and historical data for validation. On the other hand, empirical and qualitative approaches are useful for conducting ex post analyses, but it is not easy to make them operational: they seem to condemn us to wait for disasters to happen in order to advance the science. Hence, the literature presents a bifurcation between qualitative and quantitative approaches to vulnerability. This apparent disconnection between the two pictures given ex ante by quantitative approaches and ex post by qualitative approaches is a major scientific challenge. It is a problem when it comes to the definition of target populations and places for the different programs and to guide international action. Mapping and diagnostic tools tend to mobilize the same indicators or criteria beam, by transposing it from one territory to another regardless of the diversity of territories or scale effects. Also, they tend to focus on a likely snapshot, without considering that the most vulnerable individuals, social groups and place are not always the same during the unfolding of the crisis. These issues are discussed in the case of the vulnerability to floods, confronting qualitative, quantitative, French and Anglo approaches in order to enhance the robustness of methods.