Abstract: 30 releves were treated by polar ordination procedure. Four methods for construction of axis were adopted. They were: criteria of Bray & Curtis, of Yong & Lu, of Swan & Dix, and of Cottam, Goff and Whittaker. No significant difference was shown. The ordination graph obtained with the criteria of Yong & Lu, and of Bray & Curtis was more coincident with cluster analysis results. It implied that, after excluding the very irrelevant releves the only need for the establishment of the terminal stands was the variation of vegetation, no additional requirement would be necessary. But owing to avoid the terminal stands are too unrelated to rest stands, we suggest that in case those stands with 100 dissimilarity index could not be selected them as terminal stands. Thanks of the results of polar ordination were basically coincident with the cluster analysis results, it was proven that the cluster analysis can better reflect the relation between stands. Therefore, these two procedures, cluster analysis and polar ordination, are complimentary.