---

In water quality modeling, erosion and deposition of cohesive sediments are processes of paramount importance. Existing classic and processed based models can not predict the erosion flux and its trend properly. In order to evaluate the ability of the existing models, the Partheniades model as the most frequently used and Govindaraju and Winterwerp-Vankesteren Models as the most recent ones are evaluated by a set of laboratorial experiments. The sediment and water samples have been extracted from Sefidrud Dam reservoir bed. The test results have shown that despite the Partheniades model prediction, the results strictly depend on consolidation period and the erosion flux has decreasing exponential behavior with respect to time. Moreover, it has been found that the flux values are 2-7 times less than the model results. Govindaraju model is sensitive to consolidation but the temporal behavior and the quantity of the erosion flux is almost similar to Partheniades model. Winterwerp-Vankesteren model results show much less values than the latter two models but the curve behavior is the same as the latter models. It seems that these models are not able to simulate the variation of flux with time both in short term and long term. The laboratory results have shown that erosion flux has an exponential decreasing behavior with respect to time and consolidation. So if one can correlate the erosion flux with physico-mechanical parameters of bed sediments, such as Shear strength, dry density, cohesion, Gelling concentration and entrainment rate which are time and depth dependent, it will be possible to derive the more sophisticated physically and process based models.