As a result of recent floods caused by mountain rivers, more insight in the morphological processes concerning sedimentation problems and the bottom roughness to predict the expected water levels, is necessary to preclude future catastrophes. The aim of the study is to obtain more insight in the prediction of bottom roughness for situations comparable to mountain rivers. From a literature survey the importance of the relative submergence, defined as the ratio between the water depth and the bed particle size, usually written ae/D84 becomes clear. Also the Froude number would influence the flow resistance. Therefore, the research focused on these two parameters. In the Laboratory of Fluid Mechanics a number of tests has been carried out with uniform hemispheres as bed material, and varying slope, discharge and bed particle size to provide the desired range for the relative submergence and Froude number. The uniform bed particles gave the possibility to measure a uniform water depth using identical locations, defined as similar positions on a sphere at different places in the flume. The experimental results show indeed a high correlation between the Chézy value and the relative submergence, but there is no clear relation with the Froude number. If the results are compared to commonly used roughness prediction formulae (like Stickler's) it can be concluded that these do not apply for other areas than that of smallscale roughness. Besides, existing flow resistance formulae developed for mountain rivers, either determined from measurements in rivers or in models, do not highly correlate with the test results, except for Rosso's formula. Unfortunately, it is not ideal to use for mountain rivers in practice since a lot of the required parameters are difficult to obtain. Therefore a formula is proposed for the area where ae/D84 ? 3.85 that correlates highly to the experimental results and is based on Stickler's equation. For larger relative submergences the formula found by Strickler is applicable.