Accurate classification of cancer plays an important role for cancer treatment. The advancement of microarray technologies improves the accuracy of cancer diagnosis. Recently, scientists identify more informative genes from thousands of genes for accurate cancer detection. In this paper, Genetic Algorithm (GA) with bagging is developed for DeoxyriboNucleic Acid (DNA) classification. To remove the noise and data integrity, GA is applied to find the informative genes from the microarray data. It uses Backward Selection (BS), Forward Selection (FS) and Branch and Bound Selection (BBS) algorithms to select the sub-set of genes. Then bagging is employed to classify the selected genes to normal or abnormal. The evaluation of DNA classification system is performed on five cancers; colon, Central Nervous System (CNS), ovarian, leukemia and breast. Results show that the accuracy of GA-BBS with bagging algorithm is better than GA-BS and GA-FS with bagging. For all datasets, GA-BBS with bagging provides no misclassification and gives the highest performance (100%) in terms of sensitivity, accuracy and specificity. Based on results, it is concluded that ‘best’ prediction system is GA-BBS with bagging classifier.