There are many factors that influence the casualty in a earthquake,and the factors always relate each other.In this paper,some factors are selected as the evaluation indicators,such as time of earthquake、 population density、 earthquake forecast、 rate of building damage、 resistance level、earthquake magnitude、earthquake intensity.First using the method of principal component analysis to calculate the principal components,then using the method of BP neural network analysis to construct the prediction model.Test for true cases prove the model has good prediction result.