Abstract: A method of fuzzy mathematics has been applied to the recognition of seismic clusters from earthquake catalogues in a given region. A seismic cluster consists of a series of earthquakes related to each other in time and space. Every strong earthquake may be preceded by its seismic cluster. The fuzzy relative degree between two earthquakes has been determined as a function of time interval△t and distance △s by the formula:eij=1e-1t+2e-2s After calculating the degrees of relatedness of all earthquakes with MM0 in some region and taking an empirical parameter A, the seismic clusters can be recognized according to the principle of fuzzy netting.By use of seismic clusters the seismicity patterns of strong earthquakes can be recognized and studied more clearly, simply and quantitatively.This method have been applied to the recognition of seismic clusters and the study of seismicity pattern of earthquakes in North China and Southwest China. The seismic clusters of 13 large earthquakes (M6.5) have been recognized clearly, and their temporal and spatial distributions studied. The linear empirical relationships between lg△T, lgL, lgS and earthquake magnitude M have been constructed approximately, where △T, L and S are the premonitory time (duration), maximum linear dimension and the area of epicenter distribution of the seismic cluster respectively. Obviously, these relationships may be useful to earthquake prediction researches.