Abstract: This paper presents a similarity-based grid clustering algorithm (SGCA). The SGCA removes some outliers or noises in the dataset by the technique of grid and disposes of border points of clusters by the method of the threshold function of border points. The SGCA clusters by the method of similarity. Scanning the dataset only once, the SGCA can discover clusters of arbitrary shapes. The experiment results show that it can discover outliers or noises effectively and get good cluster quality. The SGCA is not only suitable for some synthetic datasets, but also it has better clustering results in some high dimensional datasets. In order to improve the efficiency of SGCA, the technique of grid cores-based is used in this paper.