Taking the CSI 300 index as a sample,a quantitative analysis was carried out reguarding the volatility of stock index return rate by using the GARCH (1,1) model under the assumption of generalized error distribution and Gaussian distribution.Then the risk was measured with conditional VaR model empirically and compared with that measured with VaR model.The result shows that GARCH-GED model can describe the characteristics of the stock index return rate perfectly and the conditional VaR model can get more accurate measures of risk in the stock market than VaR model.