In this paper, we propose the novel age estimation system with the real-coded genetic algorithm (RGA) and the neural network (NN). The age is one of important information in our living. There are a lot of studies on age estimation by the computer. However, the conventional method of the age estimation, the most of them are the studies intended for an actual age. Therefore, we pay attention to the mechanism of human age perception. The apparent age feature is extracted by the fourier transform, and the important spectrum for the age perception are selected by the RGA. The age is estimated by the 3 layered NN. It is considered that it can extract important age feature using the RGA and it can analyze the important feature area. In addition, proposed method extracts the age feature at each age. In order to show the effectiveness of the proposed method, we show the simulation examples. From the simulation results, we can confirm that the proposed method works well.