Support vector clustering (SVC) faces the same over-fitting problem as support vector machine (SVM) caused by outliers or noises. Fuzzy support vector clustering (FSVC) algorithm is presented to deal with the problem. The membership model based on k-NN is used to determine the membership value of training samples. The proposed fuzzy support vector clustering algorithm is used to determine the clusters of some benchmark data sets. Experimental results indicate that the proposed algorithm actually reduces the effect of outliers and yields better clustering quality than SVC and traditional centroid-based hierarchical clustering algorithm do.