Abstract: To monitor the quality of tobacco flavor via aroma components, a GC-MS/MS (gas chromatography-tandem mass spectrometry) method was established for simultaneously determining all aroma components in tobacco flavors. Furthermore, a quality monitoring model for tobacco flavors was developed on the basis of MSPC (multivariate statistical process control). The flavor samples were separated on a DB-WAX column after extracted with acetone at the sample/solvent ratio of 1:20 and ultrasonic extraction time of 5 min. The extracts were analyzed by GC-MS/MS under the multi-reaction monitoring (MRM) mode, and internal standard correction method was applied to semi-quantify aroma components. Statistic Hotelling T² was calculated by a PCA-based MSPC method combining with the sensory evaluation of cigarettes, and a control chart was drawn by taking 3σ as the upper control limit to assess the quality of tobacco flavors. The results showed that the LODs of the most aroma components were less than 1.0 ng/mL, the recoveries ranged from 80% to 120% with the intra-and inter-day relative standard deviations (RSDs) less than 3% and 5% respectively. This method features a high sensitivity and good accuracy. The quality monitoring model based on the analysis of all aroma components objectively reflects the intrinsic quality fluctuation of tobacco flavors and is suitable for the rapid and accurate evaluation of tobacco flavor quality.