In here we present a modified version of the RUSLE model, which can perform better than the original version at mountainous, Mediterranean catchments. A weighing coefficient is assigned at each factor in the model resulting into a reduction in the model residuals (difference between modeled and observed sediment yield). Observations are based on sediment discharge measurements conducted by the Greek Public Power Corporation (PPC). The study area includes the catchments of Venetikos, Kalamas, Arachthos and Acheloos Rivers, which are very important aquatic systems of northwestern Greece. The cross−evaluation between the originally simulated (no coefficients) and observed sediment yields attributed an initial error term value of 40 213.06t ha−1. Subsequently, the model was modified (weighing coefficients were assigned regarding the PPC measurements) and the same procedure was followed, reducing the error term value to 25 937.07t ha−1. The ambiguous reliability of the PPC measurements led to their replacement by simulations using the sediment discharge rating curves methodology. The modified RUSLE model is based on different weighing coefficients assigned regarding the rating curves estimations. Results show that the error term was moreover reduced to 5 834.78t ha−1. The final equation was implemented to each subcatchment both at an annually and inter-annually scale in order to validate the model’s reliability. Overall, the results were satisfactory.