Abstract: To improve the objectivity and scientific validity of tobacco modular formulation design by digital modular formulation, a tobacco formulation module composed of upper tobacco leaves was used as the research object. Several mathematical statistics methods were used to screen the key chemical indexes that were highly correlated with cigarettes' sensory quality, and according to the actual formulation design process, the similarity algorithm and linear programming method were used to generate a formulation that was similar to that of the target blended tobacco leaves. The results showed that: 1) Nine key chemical indexes for the formulation design of this module were screened. 2) Twenty-three alternative module formulations were optimized and screened, including 12 mandatory samples, which accounted for 67.2% of the total amount. In accordance with the principle, continuation of the use of tobacco purchased between different crop years, the expansion of the use of newly purchased tobacco was realized, which was in line with the practical requirements of the modular formulation design. 3) The contents of nine key chemical indexes of the new formulation were not significantly different from those of the target formulation, except for two sensory indexes that scored higher than those of the target formulation, the differences between the other sensory quality indexes and those of the target formulation were not significant (P > 0.05). This study shows that the digital formulation design through the use of chemical indexes with strong relation with sensory quality has better practical prospects for blending applications.