In order to make up for the deficiency of the existing evaluation method of quality stability, a new method of evaluating the quality stability of cigarette production process was developed by using statistical techniques. This method first classified the variables in the production process, and different types of variables using different methods were used to calculate the quality stability index. Moreover, different weights were determined according to both the importance of different processes and different variables in processes, and then the process using weighted average method was applied to calculate the stability of quality index. Furthermore, the process (batch) using geometric mean method was used calculate the stability of quality index. The quality stability index was divided into 4 grades as follows, more than 80 for the best, the best quality; 70～80 for good, good quality; 60～70 for qualified, the quality of qualified; 60 the following poor, needed to be improved. The results showed that the method could accurately reflect the tobacco quality level and improve the reliability and stability of batch process quality evaluation conclusion compared with the six sigma analysis method. The reason of poor stability was revealed according to different parameter indexes, and the management and control during silk production process could be significantly improved.