Cancer is most deadly human disease. According to WHO 7.6 million deaths (around 13% of all deaths) in 2008 were caused by cancer. A Cancer diagnosis can be achieved with gene expression microarray data. Microarray allows monitoring of thousands of genes of a sample simultaneously. But all the genes in gene expression data are not informative. The relevant gene selection/extraction is the main challenge in microarray data analysis. Microarray data classification is two stage process i.e. features selection and classification. Feature selection techniques are used to extract a small subset of relevant genes without degrading the performance of classifier. The classifier uses these extracted relevant genes for cancer classification. In this review paper there is a comparative study of the feature selection and classification techniques. The evaluation criteria are applied to find out the best combination of feature selection and classification technique for accurate cancer classification General TermsCancer classification