This paper discusses the methods of similarity measurement of most clustering algorithms, and taking the type of attribute as a standard of choosing similarity, it expounds the methods used to measure numerical attribute, categorical attribute and mixed a