This study assessed the uncertainty in flood impact assessment (FIA) that may be introduced by errors in moderate resolution regional and moderate resolution global Digital Elevation Models (DEM). One arc-second National Elevation Dataset (NED) and one arc-second Shuttle Radar Topography Mission (SRTM) DEMs were selected to represent moderate resolution regional and global DEMs. The relative performance for scenarios based on each of the DEMs was compared to a "control" terrain (combination of surveyed river bathymetry and a 1/3 arc-second LiDAR for floodplains)-based scenario. Furthermore, a conveyance-based DEM correction technique was applied to the DEMs for investigating the suitability of the technique on selected DEMs, and determining subsequent improvement in the FIA. The May 2010 flood on the Cumberland River near Nashville, TN, was selected as the case study. It was found that the hydraulic properties necessary to implement the selected DEM correction technique could be more readily estimated from NED compared to SRTM. However, this study also prescribed alternate methods to extract necessary hydraulic properties if the DEM quality was compromised. NED-based hydrodynamic modeling resulted in a high overestimation of the simulated flood stage, but the SRTM-based model was unable to produce any reasonable result prior to DEM correction. Nevertheless, after DEM correction, both models became stable and produced less error. Error in simulated flood consequence (i.e., total structures affected and total loss in dollars) also dropped accordingly, following the DEM correction. Therefore, application of this conveyance-based correction technique is reasonably effective on both moderate-resolution regional and global DEMs. The effectiveness of the technique on moderate resolution global DEM underscores the potential for users of remote and data-poor areas.