This paper presents an approach to incorporate time-dependent dune evolution in the determination of bed roughness coefficients applied in hydraulic models. Dune roughness is calculated by using the process-based dune evolution model of Paarlberg et al. (2009) and the empirical dune roughness predictor of Van Rijn (1984). The approach is illustrated by applying it to a river of simple geometry in the 1-D hydraulic model SOBEK for two different flood wave shapes. Calculated dune heights clearly show a dependency on rate of change in discharge with time: dunes grow to larger heights for a flood wave with a smaller rate of change. Bed roughness coefficients computed using the new approach can be up to 10% higher than roughness coefficients based on calibration, with the largest differences at low flows. As a result of this larger bed roughness, computed water depths can be up to 15% larger at low flow. The new approach helps to reduce uncertainties in bed roughness coefficients of flow models, especially for river systems with strong variations in discharge with time.