Dam failure constitutes a grave threat to human life. However, there is still a lack of systematic and comprehensive research on the loss of life (L) caused by dam break in China. From the perspective of protecting human life, a new calculation method for L occurred in dam break floods is put forward. Fourteen dam failure cases in China are selected as the basic data by three-dimensional stratified sampling, balancing spatial, vertical elevation and temporal representations, as well as considering various conditions of the dam collapse. The method includes three progressive steps: Firstly, some impact factors of loss of life (IFL) are selected by literature survey, i.e., severity of dam break flood (S F), population at risk (P R), understanding of dam break (U B), warning time (T W) and evacuation condition (E C). And the other IFL of weather during dam break (W B), dam break mode (M B), water storage (S W), building vulnerability (V B), dam break time (T B) and average distance from affected area to dam (D D) are also taken into account to get a more comprehensive consideration. According to disaster system and disaster risk, these eleven IFL are divided into four categories. Through the improved entropy method, eight key IFL are further selected out of the eleven. Secondly, four L modules are built based on four categories, which are L-causing factor module (M 1), L-prone environment module (M 2), affected body module (M 3) and rescue condition module (M 4). Eventually, by using two methods of multivariate nonlinear regression and leave-one-out cross-validation in combination with coupled four modules, the calculation method for L is established. Compared with the results of Graham method and D&M method, the result of the proposed one is much closer to the actual value and performs better in fitting effect and regional applicability. In the application, L calculation and consequence assessment are carried out in the example of Hengjiang reservoir that has already broken down. At the same time, L calculation and risk prediction are used in the analysis of Yunshan reservoir, which is under planning. The proposed method can not only be applied to estimate L and its rate (f L ) under various types of dam break conditions in China, but also provide a reliable consequence assessment and prediction approach to reduce the risk of L.