Abstract: In order to explore correlation between proximate analysis components and calorific value of biomass pellet fuel，36 kinds of biomass pellets were evaluated comprehensively and calorific value was predicted based proximate analysis.Results showed that there was a linear relationship between four indicators of proximate analysis components and a negative correlation between volatiles and other indicators.Cumulative variance of two principal components extracted was as high as 85.97%.In the two principal components，volatile component accounts for the largest share of the first principal component and ash component has the largest weight of the second principal component.Prediction model for calorific value of biomass pellet established by multivariate linear regression model showed that standard deviation of external validation was 0.185 kJ/g，and relative standard deviation（RSD）was 1.02%.Prediction model was highly reliable.