Flood routing in rivers is a mathematical procedure to determine flow hydrograph at a point in a river. One of routing methods is based on solution of Saint-Venant equations for unsteady flows. Because this method is very complex and need to more information of river, other methods with simple calculations and reasonably accurate results have been extended and give generally satisfactory results for hydrologist researchers and need to lower information of river. Muskingum model is one of these methods that accuracy in evaluation of its parameters effects on predicting flood hydrograph, especially peak rate of flow. As far as genetic algorithm is an appropriate solution to determine optimized nonlinear Muskingum coefficients, in this study we used this method to determine optimized coefficients in MATLAB and for a wide range of hydrographs getting from HEC-RAS software. Then the nonlinear Muskingum coefficients were presented as functions of river characteristics and inflow hydrograph. The model has been developed using the functions and solving of differential continuty equation with Rung-Kutta order 4. To determine the accuracy of the model, measured hydrographs of 5 floods in a reach, were compared to hydrographs computed by the model and HEC-RAS software and the results were analyzed using RMSE factor and correlation coefficient. Results indicate that there was no significant diference between the model and HEC-RAS .