After all study in the field feature selection and classifier designing it is found that Feature Selection is very important task in classification and machine learning. In this paper, the authors present a methodology to select the optimum number of features from the dataset using genetic programming. Form best classifier, they select optimum number of feature for this genetic programming life cycle is run for the fifty percent of generation. After this process, they get set of optimum features. Other feature present in classifier is replace with features exist in optimum set and this process is run till last generation.