Abstract: In order to improve the existing steady state optimization methods for technological process in tobacco primary processing, a network model was established by taking technological parameters as network nodes, influential relationships as paths and influencing degrees as path parameters on the basis of Bayesian network analysis method. The key parameters in primary processing were optimized via quantitatively analyzing the relationships between technological parameters. The established network model was validated with the data of tobacco primary processing lines in Kunming Cigarette Factory, and the results showed that the deviations between the predicted values and real values of six key parameters were all less than 1% with the average deviation of 0.52%, which was lower than the average deviation (0.85%) set by technological standards. It was indicated that the established model featured higher accuracy. This method provides a support for the further optimization of technological process of primary processing.