With more than a decade of application in the dam safety industry, the RCEM is among the most widely used methods for assessing loss of life caused by dam failures. It is founded on information from dam failures obtained from various sources and catalogued by the USBR. An important component of RCEM is the use of charts that allow estimating fatality rates based on the product of flooding depth and flow velocity (DV), as the primary indicator of “flood severity”. The flooding depth and flow velocity used for application of RCEM are typically derived from numerical hydrodynamic models, used to simulate the flooding caused by dam failures. Recent advances in that area have made the use of 2D models more accessible and widespread than ten years ago. These 2D models provide flooding depth and flow velocity at a more granular scale than what was typically available when the RCEM charts were developed. When applying those detailed results to RCEM, challenges might arise if they do not align with the sources of the RCEM charts. To investigate those challenges, this paper compares three possible ways of deriving DV from the results obtained with 2D hydraulic models: obtaining localized values for individual features, using statistical measures of DV for entire areas, and using more generalized estimates based on total flow and top width. The results of that analysis show how the approach used to derive DV can significantly influence the characterization of flood severity and the corresponding estimation of fatality rates and loss of life resulting from dam failures.