Abstract: To investigate the effects of material composition on the combustion characteristics of cigar tobacco leaves by thermal analysis, the material composition and combustion characteristic parameters of cigar tobacco leaves were determined using thermal analysis techniques. Constituents significantly correlated with the combustion characteristic parameters were screened through correlation analysis, and prediction models for these parameters were established using partial least squares (PLS) regression. The results indicated that: 1)The concentration of burnout temperature (T_t), average combustion rate (R_a), and comprehensive combustion characteristic index (Z) in wrapper leaves was significantly higher than that in filler and binder leaves. 2)The peak temperature (T_max) exhibited a highly significant negative correlation with the potassium-to-chlorine ratio, total sugars, reducing sugars, and potassium content. The residual mass from thermal analysis (M_t) showed a highly significant negative correlation with hydrogen and carbon content, but a highly significant positive correlation with potassium content.The comprehensive combustion characteristic index (Z) demonstrated a highly significant negative correlation with ash and chlorine content, whereas it showed a highly significant positive correlation with total sugars, reducing sugars, oxygen content, and the potassium-to-chlorine ratio. 3)The model for peak temperature (T_max) exhibited relatively weaker goodness-of-fit and predictive capability, while the models for residual mass from thermal analysis (M_t) and the comprehensive combustion characteristic index (Z) both demonstrated satisfactory goodness-of-fit and predictive ability.