This study was conducted to evaluate the prediction of sensory attributes of tobacco smoke by the use of volatile oil components of tobacco. For analytical and sensory evaluations, twelve aging tobaccos (i.e., 3 crop years; 1998, 1999, 2000, and 4 stalk positions) were prepared in flue-cured and burley tobaccos. 61 volatile oil components and 5 sensory attributes such as irritation, impact, after taste, bitter and green were investigated. Irritation of flue-cured tobacco, and irritation and impact of burley tobacco were significantly increased with the ascending stalk position, whereas after taste of burley tobacco was significantly decreased. Significantly positive correlations among irritation, impact, and bitter were observed in flue-cured tobacco. A significantly positive correlation between irritation and bitter was observed, significantly negative correlations between after taste and irritation and between after taste and impact were observed in burley tobacco. Except for green of burley tobacco, all probabilities of multiple linear regression equations between volatile oil components of tobacco and sensory attributes of tobacco smoke were significant(P$leq$0.05). This study suggests that the multiple linear regression equations may be useful to predict the sensory attributes of tobacco smoke with a few selected volatile oil components of tobacco.