Abstract: It is aimed in this paper that "RSWA" s characteristics in predicting typhoon paths are further revealed by bulk-sample-basis computation and analysis. Statistic analysis shows that reliability of its results depends to a large extent upon the initial position of typhoon and local features of its early path(e. g. direction and speed of motion). In this way, what is imperfect with "RSWA" is fornd, making it convenient for improvement in predicting accuracy in future.In the paper, characteristics of the model are discussed in terms of prediction error only. Results presented here indicate that "RSWA" can be relyed on to a certain degree in the prediction of typhoon path, particularly for Group C typhoons, having more advantages over the olc scheme.Analysis shows that causes leading to error in a prediction method should be made up of error due to model (EM) and error due to factors concerned with the motion of typhoon (ET), the latter being a continuous function of typhoon's initial location and motion, intensity and season of occurrence.