Abstract: Coalbed mathane content was an important parameter to determine the development effect of coalbed methane. Accurate determination of the content w as a key problem for coalbed methane exploration and development research. Taking east Qinnan Block in southeast Qinshui basin as the target area, the relationship between coalbed mathane content and logging parameters were analyzed, based on gas desorption experiment of mathane content and coalbed methane drilling loggin g parameters. Six parameters were chosen as the basic characteristic quantities of BP artificial neural network prediction model, such as logarithm of effective buried d epth, volume density, spontaneous potential, ratio of deep lateral resistivity and shallow lateral resistivity, logarithm of microspherically focused resistivity, ratio of acous tic time difference and the product of natural gamma ray and compensated neutron. The BP artificial neural network prediction model for coalbed methane content was built. The model' error and results contrast of its application were analyzed. The results showed that BP artificial neural network prediction model based on logging para meters had strong nonlinear approximation ability and could actually reflect the nonlinear relationship between coalbed methane content and logging parameters. The e rror between predicted results and measured value was small, with the relative error generally less than 10%. There would be a good application prospect in using loggi ng curves to predict gas content of coal seam.