In this paper, survey on biclustering approaches for Gene Expression Data (GED) is carried out. Some of the issues are Correlation, Class discovery, Coherent biclusters and coregulated biclusters. Each table entry is called an expression value and reflects the behaviour of the gene in a row in the situation in column. The goal of biclustering is to identify "homogeneous" submatrices. Given a gene expression data matrix D=G*C= {dij} (here € [1, n], j €[1, m]) is a real-valued n?m matrix, here G is a set of n genes {g1, g2.., gn}, C a set of m biological conditions {c1, c2..,cn}.