Nowadays, the flight distance of champion team of the birdman rally contest was over 400 m. In order to break the record, we have to consider the weather conditions, flight conditions and design optimization of the glider. For this purpose, we developed the three-dimensional flight simulator and the optimization program based on genetic algorithm. The simulation result of the three-dimensional flight simulator was reasonably agreement with the real flight. GA (Genetic Algorithm) optimization was carried out with two hundreds samples and fifty generations. The dominant genes of airplane for the contest were obtained with GA. The performance map showed the optimal airplanes such as the champion team were sensitive and not easy-handle. The simulator was suggested another optimized model with flexible wing that was not sensitive to flight conditions. The proposed model got the fourth prize and the referee's award of the birdman rally contest 2005.