The numerical 1-D HEC-RAS modelling tool was supported by the estimation of the sieve curve changes procedure to measure the scale of predicted discharges along a stretch of stream in southern Poland on the Olkusz Upland. The procedure was calibrated in southern Poland on the mountain streams during high-stage events, using a radiotracer application in bedload transport. Particular terrain hypsometry, created by the dissolution of limestone, forced the deep erosion of the river valley bottom; it is here that the current shape of the riverbed of the Prądnik stream is placed. While numerical modelling is widely used in hydraulics, standards have been set for the estimation of flood risk zones; these estimations suggest that the densities of the measured cross-sections are less then optimal, and that the erosive processes are more frequent. This was proved by identifying a number of erosive sections. A new procedure proposed combining the prediction of grain size distribution with hydraulic modelling. Calculations using the estimation of sieve curves, based on the processes of creation and destruction in the armouring layer, have proven to be a challenge for the existing standards of hydraulic modelling. We believe that it is easy to expand the usefulness of the 1D model by utilising its results for this procedure. For the purpose of this type of analysis, dense cross-section measurements are involved, careful modelling is required and a wide range of additional in-field data has to be gathered. For the interpretation of the results, the relation between channel-forming discharge, bankfull discharge, present and critical shear stresses, as well as the mean diameter of the grain size and other estimated sieve curve parameters, were evaluated. Channel-forming discharge is smaller than the bankfull discharge in more than one third of the segment where the erosion process is more frequent and the stability of the riverbed is compromised. Channel-forming discharge was at least twice as high in the stable sections, compared to the erosive section. The presented method will help to find unstable riverbed sections, in order to mitigate the dimension of river training techniques and protect the natural state of the river. While we are in the period of development in this region of Europe, limiting the scope of interference in rivers and streams by applying this method may create an opportunity for the concept of river training close to nature.