Abstract: Clustering is a very active research branch in data mining field.The research about the arbitrary shape clustering is an open problem.In this paper an inter-cluster dissimilarity measure taking into account the frequency information of the categorical attribute values is introduced.An arbitrary shape clustering algorithm is proposed by defining the similarity degree between an object and a cluster.It can be used for the mixed attributes dataset.The experimental results on the synthetic and real-life datasets show that the proposed algorithm is feasible and effective comparing to other classical algorithms.