Abstract: The paper aims to study the optimization of the key processing parameters for microwave expansion tobacco stem during cutting process. Tobacco stem blending modules (upper tobacco stem blending module was U, middle tobacco stem blending module was M and lower tobacco stem blending module was L) of a brand of cigarette were selected as study objects. Uniform experimental design was adopted for the seven key processing parameters during microwave expansion tobacco stem during cutting process, BP neural networks models of the key processing parameters during microwave treatments process of tobacco stem were established, and the models were used for simulation and optimization of the key processing parameters during microwave treatments process of tobacco stem. Three grades of sensory evaluation score: high, medium and low, and the corresponding technology combination for module U, module M and module L from the simulation results were chosen. Pilot production and sensory taste of microwave expansion cut stem were completed according to the chosen process combination. Microwave expansion cut stem of pilot production was chosen and well mixed by mixture tests design, and substituted for the cut stems made by conventional process with the same proportion in the cigarette of class C. Compared with the certain brand of cigarette, the key processing parameters and ratio of module U, module M and module L during microwave treatments process were verified by sensory evaluation.