Abstract: Dynamic prediction of gas emission based on multi-source information fusion was a kind of new mine gas emission prediction method with a combination of traditional mine gas emission prediction and modern computer programming. Through the multi-source information fusion, such as mine measured gas content, geological exploration gas content, K₁-p or △h₂-p curve, the inversion coal gas content of coal roadway driving gas emission and so on, it obtained the gas occurrence regularity of coal seam and more accurate gas content distribution. Combined with the gas content distribution and sub source prediction method, it built the mathematical model of coal seam gas content and gas emission under the same mining condition. It used the new working face gas emission data and mine statistical method to track and correct the mathematical model of gas emission constantly, and formed the mathematical model after fusion, thus achieved the target tracking for mining area gas emission and the dynamic prediction of unmining area gas emission.