Based on the GPD-Copula-CoVaR model, this paper investigates the dynamic correlation and risk spillover effect between real estate and banking, over the period 2000 M01-2017M06 in China. In addition, we examine the forecast effect through Monte Carlo method. The results show that the correlation between real estate and banking is highly related to the prosperity and political policies of the real estate market, and the correlation is higher when market in slump. The GPD-BB1 Copula model is suit for the risk situation. The out-of-sample forecast shows that GPD-BB1 Copula model coverage of the actual risk loss is higher, so the prediction ability is better. In addition, the two marks have bidirectional risk spillover effects in extreme situation, and real estate on the banking's spillover is stronger, about 40%. This article provides an effective method for measuring the risks associated with banking and real estate in favor of authorities for risk management.