GC and GC/MS were used to analyze the volatile substances in 100 Maotai-flavor liquor samples with 5 qualities from a liquor company. Combined with odor active values and chemometrics， the identification model was established and the key difference substances were found. The results showed that a total of 29 volatile components were detected by direct injection method， among which 21 compounds contributed to the formation of Maotai-flavor liquor aroma. Clustering heat map can intuitively show the content differences of volatile components and the clustering process in wine samples of different quality. Principal components analysis was used to construct the identification model， and the results were consistent with the cluster heat map analysis. Partial least squares discriminant analysis was used to identify 12 key differential components in five different qualities， including ethyl 2-methylbutyrate， ethyl hexadecanoate， 1-butanol， 1-propanol， 1， 2-propanediol， 3-methylbutanoic acid， 2-methylpropionic acid， acetal， acetaldehyde， furfural， furfuryl alcohol， phenylethanol. By substituting the quantitative analysis results of different grades of Maotai-flavor liquor from other three liquor companies into the established identification model， effective discrimination can be achieved.