Feature selection is an important issue in flight data mining. By selecting only relevant features of flight data, higher prediction accuracy can be expected and computational complexity can be reduced. In this paper we propose a novel two-phase flight data feature selection approach using both filter and wrapper. It begins by running artificial neural network weight analysis (ANNWA) as a fi1ter approach to remove irrelevant features, then it runs genetic algorithm as a wrapper approach to remove redundant or useless features. We demonstrate the usefulness of the proposed approach on two real-world datasets based on flight data. Our algorithm reduces the size of flight data feature space significantly without compromising the classification or the prediction performance.