Abstract: To obtain information of characteristic components of tobacco flavors for casing and realize identification of monomer formulas for blended fragrance bases for casing, an untargeted metabolomics characterization method for natural flavors for casing based on liquid chromatography-tandem mass spectrometry (LC-MS/MS) was developed. A qualitative and quantitative data matrix covering 1 194 metabolites from 47 monomer samples of natural flavors was constructed. An algorithm for monomer formula identification of blended fragrance bases was developed on the basis of this metabolomics data matrix, which realized flavor discrimination and imitation replacement of commercial fragrance bases. The results showed that: 1) The developed metabolomics data analysis algorithm can effectively identify monomer formulas in blended fragrance bases, with the qualitative identification accuracy of monomers ranging from 70% to 85% (the criterion was that the possibility ranking of a certain monomer existing in the fragrance base reached top 10 out of 47 candidates). 2) The metabolic characteristics of different monomer flavors differed significantly, and the recognition sensitivity of the blended fragrance base formula analysis algorithm varied greatly among different monomer flavors. 3) The metabolomics data analysis algorithm for blended fragrance bases exhibited higher sensitivity to various tobacco extracts, which could be accurately identified even at extremely low contents (mass fractions), presumably due to the strong characteristics of the metabolomics data of tobacco extracts. 4) Combined with professional perfumers’ experiences, flavor discrimination and imitation replacement experiments were carried out on two commercial tobacco fragrance bases for casing. By comparing the flavor profile and cigarette flavoring performance of imitated fragrance bases with commercial ones, the imitation replacement of the two commercial fragrance bases was successfully achieved.