Abstract: In view of the fact that CO, C₂H₄ and C₂H₂ can not satisfy early prediction of spontaneous combustion in coal seam, the paper, based on the B₂ coal seam spontaneous combustion gas product test data in the Jiangou Coal Mine of Shenhua Xinjiang Energy Co., Ltd., selects the oxidation product concentration of coal sample at 100 ℃ to 300 ℃ and uses the grey correlation method to calculate and analyze the coal sample and find correlation between the gas concentration and the combustion temperature, and to order the correlation according the calculated results, and determine reasonable indicator gas for coal seam spontaneous combustion. The calculation results show that the correlation degree of γ3>γ5>γ4, that is,φ(C₃H₈)/φ(CH₄), φ(CO)/φ(CO₂) and φ(C₃H₈)/φ(C₂H₆) are most closely related to the coal combustion temperature. The correlation of three indicator gases is more than 0.6, and compared with φ(C2H4)/φ(C₂H₆), the selection of the three indicators is rationality. Therefore, φ(C₃H₈)/φ(CH₄),φ(CO)/φ(CO₂) and φ(C₃H₈)/φ(C₂H₆) can as indicator gases for early prediction of B₂ coal seam spontaneous combustion in the Jiangou Coal Mine.