Gene selection is one of the research issues for improving classification of microarray gene expression data. In this paper, a gene selection algorithm, which is based on the modified Recursive Feature Elimination (RFE) method, is integrated with a Support Vector Machine (SVM) to build a hybrid SVM-RFE model for cancer classification. The proposed model operates with a two-stage gene elimination scheme for finding a subset of expressed genes that indicate a disease. The effectiveness of the proposed model is evaluated using a multi-class lung cancer problem. The results show that the proposed SVM-RFE model is able to perform well with high classification accuracy rates.