Miscible and/or near miscible gas injection processes are those among the most widely used enhanced oil recovery techniques. And, a CO2-oil miscible flooding is more effective to enhance the local displacement efficiency. The successful design and implementation of a miscible CO2 injection project is dependent upon the accurate determination of MMP（minimum miscibility pressure）, the pressure above which the displacement process becomes multicontact miscible. This paper considers the multicontact process during the miscible flooding, and presents a method to get the characteristics curve of multicontact, and analyzes the character of characteristics curve for CO2 flooding, and combines some main factor (reservoir temperature, oil composition) that may infect MMP,then, we suggest a correlation to make a prediction for the minimum value of characteristics curve of multicontact. Based on the work above, we suggest a new method to predict MMP for CO2 flooding. Our method needn’t locate the key tie lines, and avoid the unfavorable factors existing in the MOC and the mixing-cell method which leads to the failure of MMP prediction, so as to overcome the shortcoming of traditional correlation which didn’t consider the process of multicontact. For all of these reasons, our method becomes more practical and more accurate. The prediction error for our method is within 8%.