The evaluation of major S&T program is of significance for implementing the national S&T plans. However, with the complexity of major S&T program, the evaluation would be faced with a lot of problems. This paper structures and perfects evaluation index system based on balanced scorecard, and explains the relationship of evaluation index briefly. Then, the evaluation index carries out factor analysis with primary component analysis method. Thus, the primary components are obtained. Subsequently, the evaluation model is constructed via GA-BP neural network. Finally, taking provincial S&T program of Tianjin city as the case, the paper proves that the model is not only scientific but also practical.