To identify signature genes for the tumor progression of lung squamous cell carcinoma, which provides a deeper theoretical basis for further explanation of its inherent mechanism, new targeted drugs and treatments development. The pattern recognition method was used to analysis the genome-wide mRNA gene expression (GE) values, methylation values (ME), and copy number variation (CNV) data. To overcome the disadvantages inherent in the genome-wide data such as ultrahigh-dimensional-small-size, high-noise and multi-correlation among genes, and to overcome the predominate influence of the whole genome to the dozens of signature genes, a new iterative multiple variable selection strategy was used to identify signature genes step by step. The importance of genes was comprehensively evaluated by their significant difference with SAM (significant analysis of microarray), statistical analysis using PLS (partial least squares), known biological functions and contributions to the classification model. 67 GE signature genes, 70 ME signature genes and 31 CNV signature genes were identified from the LUSC stageⅠ～Ⅲ patient samples in TCGA (The Cancer Genome Atlas project) database. The corresponding accuracies from 5 fold cross-validation are: 86.29%, 90.92 % and 69.16% respectively. The genetic network analysis and pathway analysis using KEGG (Kyoto Encyclopedia of Genes and Genomes) and IPA (Ingenuity Pathway Analysis) indicated the highly related relationship among these three kinds of genes. They also indicated the immediate relationship between our signature genes and the progression of LUSC which is very important to the understanding of its mechanism and to the development of new targeted therapy.