Abstract The purpose of study was to improve the three-staged skills of games analysis in Table Tennis with Rasch Measurement. The subjects were three students who participated in three matches of round robin tournaments of the recruitment of Table Tennis in university. The methods were compared by the games analysis between new and old method of three-staged skills in Table Tennis. The old method used data-collecting and statistic analysis to decide final assessment by subjective index. The new method was based on the model of many-facet Rasch measurement. The study design was divided into the master team, guest team and three stage skills and etc., which were as the three facets. The many-facet Rasch measurement model was i'mnnmi0nmi1I-B-BPPlog . The followings were results: 1. Wu and Lee’s traditional method’s evaluation (1990) was uncertain. 2. The new method of Rasch measurement had good reliability, validity and the representative of samples. 3. Compared with two methods from summaries, there were eight contents being good with new method. The conclusion of this study was that the Rasch measurement had theoretical, objective and precise, and a better method for the three-staged skills of games analysis in Table Tennis.