Conventional seismic attributes can be effectively used in 4D seismic monitoring. However, these seismic attributes have limitations for the identification of reservoir heterogeneity or anisotropy, and could not explain the characteristics of reservoir changes in the process of CO2 injection and storage. Due to the heterogeneity of the reservoir, the reservoir property has not been interpreted clearly during CO2-EOR and storage in G89 block of Sinopec Shengli Oil Company,China. Based on gray level cooccurrence matric, we applied seismic texture analysis method into 4D seismic data to discriminate CO2 migration pathways and distribution. The difficulty of texture attributes analysis of 4D seismic difference lies in attributes selection firstly. That means which attribute can be chosen and used to identify the changes of reservoir parameters before and after CO2 injection. And then how to obtain the 4D seismic difference attributes of the reservoir. We calculate the seismic texture attributes of the reservoir by means of the caprock consistency processing method of two vintages of 3D seismic datasets. Then the difference attributes background interference is reduced much, and the contrast between the fault and background information is increased, which obviously enhances the visibility of the plane characterization of the reservoir. The difference of 3D seismic texture attributes shows the better image of breakthrough path and distribution of CO2. It indicates the dominate channel lead to the CO2 breakthrough and moved to the highest location of the geological structure