Abstract: A data noise reduction algorithm based on the linear Kalman filter was developed to solve the problems of process noise and measurement noise at the time of data acquisition of instantaneous flavoring flow rate and moisture content in output tobacco during tobacco flavoring. It extracts signals from the noises via the values of status and measurement values and solves linear filtering problems by iterative computation. The developed algorithm was tested on the cigarette brand "Yuxi (Soft)" in Yuxi Cigarette Factory. The results showed that at the steady stage of flavoring, the standard deviation and coefficient of variation of the instantaneous flavoring flow rate were reduced by 54.83% and 54.84%, and those of the moisture content in the output tobacco were reduced by 51.15% and 51.15%, respectively, moreover, the over-regulation of the instantaneous flavoring flow rate and moisture content in the output tobacco decreased at the initial and final stages and at the same time their smoothness was also promoted. This method helps to improve the reliability and stability of the tobacco flavoring cylinder control system.