Abstract:This paper discusses how to use the similarity coefficients, i.e. associationcoefficients, to analyse the floristic resemblance and to resolve problems in comparative study of numerous floras being short of disirable comparison, and in floristic affinity tending to 'phenology" similarity by using improper formula, or neglecting the negative matching shared numbers and so on. The authors take flora similarity in two separate circumstances, narrow and broad similarities, the former points to the normal matching shared numbers of any two floras, and the later, to both the negative and normal matching shared numbers of any two floras. The floristic resemblances, similar relationships and the substance of similarity coefficients from the angle of Set-Theory are discussed. It is considered that the formula of similarity coefficient set up by P. F. Russell and T. R. Rao(1940) is suitable for discussion and determination of narrow similarity, but the simple matching coefficient put forward by R. R. Sokal and C. D. Michener(1958) is suitable for broad similarity. Among them, the formula of similarity coefficient by P. F. Russell and T. R. Rao (1940) is a special form i.e d=0 of the formula of simple matching coefficient by R. R. Sokal and C, D. Michener(1958). The authors suggest two new concepts, one is the similarity in population, and the other, similarity in areal-type, so as to have a unified study of the noristic similarity in different regions separately according to the study of flora in china by Prof. C. Y. wu. In this paper, six floras are used as examples.