Feature selection is an important data preprocessing technique and has been widely studied in data mining, machine learning and granular computing. In this paper, we introduced an effective feature selection method using the hybrid approaches, that is, use the mutual information to select the candidate feature set, then, obtain the super-reduct space from the candidate feature set by a wrapper search approach, finally, the wrapper method determined to select the proper features. The experimental results show that our approach owned the obvious merits in the aspect of classification accuracy ratio and number features selected by extensive comparing with other methods.