The effective solution to sparse linear systems which is one of the hotpots in numerical computation focuses also on the preprocessing arts for the systems. From the view of classification to the techniques,this paper sums up the preprocessing methods. Concretely,it introduces the fillreducing strategies to decrease the storage and keep the matrices sparse in the solving process,the matching strategies for  different structural kinds of matrices to obtain  different diagonal dominances, the sparse approximate inverse proconditioner holding the parallel characteristics in nature,and the parallel methods for incomplete factorization,and so on.