This paper mainly discuss the BP neural networks based prediction and evaluation method. In order to avoid the drawbacks of the classical BP networks, the PCA method is introduced, based on which a combination method called the PCA based BP neural networks is proposed. Finally, an experiment on evaluation for the urban management is given. The results demonstrate the validity and feasibility of the PCA based BP neural networks. And the comparison shows that: the PCA based neural networks does have a better performance than the classical BP neural networks.