Hybrid water-sediment flow models for river reaches have been for predicting sediment and sediment related pollutions in water courses. The models are developed by combining sediment rating model and the Muskingum model applicable for a reach. The models incorporate sediment concentration and water discharge variables for a river reach; allow defining downstream sediment rating curve in terms of upstream water discharges. The model is useful in generating sediment concentration graph for a station having no water discharge records. The hybrid models provide forecasting forms that can be used to forecast downstream sediment concentration/water discharges 2kx time unit ahead. The forecasting models are useful for applications in real time namely, in the real time management of sediment related pollution in water courses and in issuing flood warning. Integration of sediment rating model and the Muskingum model increases model parameters and nonlinearity requiring efficient estimation technique for parameter identification. To identify parameters in the hybrid models genetic algorithm (GA) based optimization technique can be used. The new model relies on the Muskingum model, obey continuity requirement and the parameters can be used in the Muskingum model with water discharges to estimate/predict downstream water discharge values. The proposed model formulations are demonstrated for simulating and forecasting sediment concentration and water discharges in the Mississippi River Basin, USA. Model parameters are estimated using non-dominated sorting Genetic Algorithm II (NSGA-II). Model results show satisfactory model performances.