Abstract: Partial least squares regression (PLSR) was employed to analyze the correlation of tobacco chemical components and sensory quality. The results showed that sugar/alkaloids and sugar/nitrogen had positive correlations with comfortable and aftertaste attributes, and negative correlations with irritation and off-taste attributes. Additionally, alkaline composition including total alkaloids and volatile alkali had significant negative correlations with aroma quality attributes (p<0.05). Total nitrogen, total alkaloids, protein, volatile alkali and petroleum ether extracts contributed significantly to the increase of irritation, while total sugars and reducing sugars had significant effect on masking harsh-taste (p<0.05). K, Cl and volatile acids contributed to the increase of the sensory comfortable attribute, and the reduction of sensory impact (p<0.05). PLSR solved the multiple linear regression problems in which the use of traditional multivariate methods is severely limited, such as limited number of observations and collinearity. The results of this study are important for guiding cigarette tobacco processing and modifying technology.