Abstract: Objective: To screen for and identify metastasis-related genes in breast cancer by comparing oncogene expression maps of the primary breast cancer and matched lymph node metastases, in order to discuss the clinical significance for molecular grouping of the breast cancer patients with lymph node metastasis. Methods: cRNA amplification was used to label primary tumor specimens with Cy3 and lymph node metastases with Cy5. Paired samples were combined to hybridize to oligonucleotide microarrays with 21,329 functional human genes. The 2.0-fold or plus genes with the same differential-expression (up-regulated or down-regulated) potential were found in at least 16 of the 30 pairs of samples. Sixteen of the 30 sample pairs had genes differentially expressed at least 2.0-fold. Non-supervised cluster analysis (i.e. hierarchical cluster analysis) was performed for the specimens using Cluster3.0. Results: A total of 94 differentially expressed genes were identified, among which 41 showed upregulated expression and 53 showed downregulated expression in metastatic tissues. The genes were related to cell migration and adhesion, extracellular matrix , proteinase activity, signal transduction, transcription and regulation , as well as cell growth and apoptosis. Based on cluster analysis of the total gene expression maps, primary cancer in 83.3% (25/30) of the cases and the paired lymph node metastatic carcinomas grouped into clusters. The 30 cases were divided into 2 groups based on the cluster analysis of the differentially expressed genes, in which most of the cases in group A had poor prognosis with T_3-4 tumors (P<0.05), were estrogen receptor negative (P<0.05), had histological grade- Ⅲ (P=0.073) and had distant metastasis (P=0.085), compared to group B. Conclusions: The genes differentially expressed between primary breast cancer and lymph node metastasis hold the key to metastasis. Primary cancer and metastatic carcinoma are classified together in our cluster analysis, suggesting that lymph node metastatic carcinoma is a subclone of the cells with high metastatic potential from the primary cancer, thus having a similar gene expression map where the differentially expressed genes supply the necessary components for the metastatic phenotype. Based on the expression of those differentially expressed genes in the primary tumors, molecular grouping can be conducted for the cases of breast cancer with positive lymph node metastasis and the prognosis can be predicted.