The volatility of stock market is closely related to the real economic trend and macroeconomic policies． This paper analyzes the effect of economic policy uncertainty ( EPU) on the SSEC volatility using GAＲCH-MIDAS model． Out-of-sample volatility forecasting comparison is also made between the GAＲCH-MIDAS model and several commonly used GAＲCH-type models． The empirical results show that the EPU can well explain the long term section of stock market volatility and significantly improve the forecasting accuracy of SEEC volatility． Meanwhile，the model confidence set ( MCS) test confirms that the forecasting result of the GAＲCH-MIDAS model based on mixed-frequency data outperforms that of many other commonly used GAＲCH-type models．