Abstract: In this paper, the principle and method of Fuzzy Clustering is introduced on the base of the fuzzy theory. The flood disaster degree of "96.7"floods in Xinjiang, which is one of the serous flood disasters taken place inXinjiang, is classified by using Fuzzy Clustering method. In the application of Fuzzy clustering analysis, disasterareas are taken as the clustering objects X = {X 1, X 2 … Xm } combining those regions, which were damaged inthe "96.7" floods. Taking some essential factors (the injured population, damage areas, destroyed houses, di-rect economy losses) of disaster losses as the clustering quotas (Xi 1;. Xi 2; … Xin), and then the floods disaster hasbeen divided into five grades (K=1, 2, 3、4, 5) according to the level of economy deveiopment and populationdensity of Xinjiang. Different grades of the disaster regions in the"96.7"floods are made out by using fuzzyclustering which deterrmined by different coefficient"A". The results are as follows: (1) The extreme heavydisaster areas are: Changi, Turpan, Bayingolin prefecture and Xinjiang Production and Construction Groups (state-running farms) ; (2) The heavy disaster areas are: Urumqi city, Tacheng and Kasher prefecture: (3) The middle disaster areas are:Bortala, Hami and Kizilsu prefecture. The classification results are mostly corre-sponding to the flood disaster losses of the "96.7" floods. All mentioned above indicates that the fuzzy clusteringis an effective method to deal with the classification of disaster grade.

Key words: Fuzzy clustering, Flood disaster, Xinjiang