A set of laboratory experiments on bare, rough soil surfaces was carried out to study the relationship between soil surface roughness and its hydraulic resistance. Existing models relating roughness coefficients to a measure of surface roughness did not predict the hydraulic resistance well for these surfaces. Therefore, a new model is developed to predict the hydraulic resistance of the surface, based on detailed surface roughness data. Roughness profiles perpendicular to the flow are used to calculate the wet cross-sectional area and hydraulic radius given a certain water level. The algorithm of Savat is then applied to calculate the hydraulic resistance. The value for the equivalent roughness, which is used in the algorithm of Savat, could be predicted from the roughness profiles. Here, the tortuosity of the submerged part of the surface was used, which means that the calculated roughness depends on flow depth. The roughness increased with discharge, due to the fact that rougher parts of the surface became submerged at higher discharges. Therefore, a single measure of surface roughness (e.g. random roughness) is not sufficient to predict the hydraulic resistance. The proposed model allows the extension of the flow over the surface with increasing discharge to be taken into account, as well as the roughness within the submerged part of the surface. Therefore, the model is able to predict flow velocities reasonably well from discharge and roughness data only. Copyright © 2000 John Wiley & Sons, Ltd.