Abstract: Coalbed methane (CBM) is a kind of gas reservoir which mainly exists in coal reservoirs in adsorption state. The prediction of coalbed methane enrichment area by seismic exploration is an important method to reduce the exploration risk and improve the recovery of coal-bed methane wells. For the reason that the occurrence mechanism of coalbed methane is completely different from that of conventional oil and gas, and the seismic petrophysical mechanism of coalbed methane occurrence is still unclear, conventional oil and gas prediction methods are often ineffective when applied to coalbed methane prediction. In this study, an evaluation method of coalbed methane enrichment unit is proposed, which takes the prediction of geological elements of coalbed methane occurrence as the research object and integrates multiple seismic attribute parameters. Setting the Wuxiang South Block in Qinshui Basin as the study area, the thickness of coal seams and the lithology of roof and floor were predicted basing on the physical analysis of coal seams and surrounding rock and wave impedance inversion,. The sedimentary microfacies were analyzed using waveform clustering analysis method before and after coal accumulation, which led to the significant difference of the density of coal rocks with different coal body structures, thus the den-sity attribute was used to predict the plane distribution of tectonic coal. Since the fracture density has a good correlation with the coal seam permeability, it is used to predict the coal seam permeability. The forward modeling results show that the attribute of λρ and μρ is sensitive to the content of adsorbed gas in coal seams, therefore it is selected to predict the adsorption gas content. Relying on these abovementioned five geological parameters: the lithology of the coal seam roof and floor at the well point, the sedimentary microfacies before and after coal accumulation, the coal body structure, the permeability and the gas bearing property as samples , the BP neural network model of the coalbed methane enrichment area is constructed for the known coal seam gas content at the well point,, which presents as the building of the nonlinear mapping relationship between the five types of geological elements and the enrichment area through sample training, which results in the prediction of coalbed methane enrichment area . The results show that the coalbed methane enrichment predicted by multi geological parameter fusion method has a high coincidence rate with current drilling, which reduces the multi solution of single factor prediction.