The forecasting model of energy crops in anaerobic digestion was established based on feedstock component by using linear regression method. Taking brazil pennisetum purpureum, south pennisetum purpureum, dwarf pennisetum purpureum, Taiwan pasturage B and seven types of hybrid pennisetums harvested in different months as samples; the component of C, N, C/N, cellulose, hemicellulose, and lignin as independent variables; and the cumulative methane yield of energy crops as dependent variable. Linear regression find that methane yield have weak correlation against C and hemicellulose contents (R2 = 0.02, R2 = 0.03), certain correlation against C/N ratio and cellulose content (R2 = 0.37, R2 = 0.313), and strong correlation against N and lignin contents (R2 = 0.461, R2 = 0.51). By multiple linear regression analysis, we obtain two forecasting models which are high degree of confidence, significant correlation and small error (R2 = 0.779, R2 = 0.783). By curve fitting analysis and standard error calculation, the model is proved to be reliable.