Riverbed armoring in gravel-sand rivers involves complex interactions among suspended load, bed load, and bed material, forming a non-steady “scour–armoring–exchange” process. Traditional prediction methods often oversimplify these mechanisms and neglect sediment exchange, leading to limited accuracy and applicability. This study aims to develop a new riverbed armoring calculation method that explicitly incorporates sediment exchange between suspended load, bed load, and bed material to improve prediction performance. Four representative domestic and international riverbed armoring models were first compared under different hydraulic and sediment conditions to identify their limitations. Based on sediment transport statistical theory and an improved active-layer sediment mass conservation equation, a new method was proposed that considers the coupled “suspended load–bed load–bed material” exchange process. The model was validated against both flume experiments and field data from the downstream reaches of the Danjiangkou and Three Gorges Reservoirs. The results show that compared with the traditional methods, the proposed method significantly improves the prediction accuracy. The calculation accuracy of the new method is considerably higher than that of conventional approaches in both natural rivers and flume experiments. When applied to natural rivers, the RMSE and MAE values are 4.9% and 3.3%, respectively, while for flume tests they are 5.5% and 2.5%, respectively. It effectively reproduced the dynamic feedback between flow scour and bed armoring, accurately simulating both the gradation of the armoring layer and the depth of bed scour. Sensitivity analyses demonstrated that dynamic variation of the active-layer thickness and appropriate definition of armoring stability are essential for accurate modeling. The newly developed method, grounded in sediment transport statistics and active-layer balance theory, provides a physically-based and reliable approach for predicting armoring in gravel-sand beds. It addresses key limitations of conventional models by coupling multiple sediment transport modes and dynamic bed evolution. The results enhance understanding of riverbed armoring mechanisms and offer a robust tool for forecasting downstream channel adjustment below large dams.