The successful development and application of a conceptual rainfall-runoff model depends mainly on how well it is calibrated. This paper introduces three optimization methods, namely Rosenbrock, Simplex and Genetic algorithm, for automatic calibration of the hydrological models, The performances of the methods were analyzed on the basis of the application of them to the Xinanjiang model using 14 catchments data. The effects of the parameter values of the Genetic method on calibration of the model parameter values were preliminarily investigated. The results suggest that the Genetic algorithm with further tuning by other two methods can provide an efficient and robust means for calibrating the conceptual rainfall-runoff models.