Shale gas is becoming an important component of the global energy supply,with permeability a critical controlling factor for long-term gas production.Obvious deviation may exist between helium permeability determined using small pressure gradient (SPG) methods and methane permeability obtained under actual field production with variable pressure gradients (VPG).In order to more accurately evaluate the matrix permeability of shale,a VPG method using real gas (rather than He) is established to render permeability measurements that are more representative of reservoir conditions and hence response.Dynamic methane production experiments were performed to measure permeability using the annular space in the shale cores.For each production stage,boundary pressure is maintained at a constant and the gas production with time is measured on the basis of volume change history in the measuring pump.A mathematical model explicitly accommodating gas desorption uses pseudo-pressure and pseudo-time to accommodate the effects of variations in pressuredependent PVT parameters.Analytical and semi-analytical solutions to the model are obtained and discussed.These provide a convenient approach to estimate radial permeability in the core by nonlinear fitting to match the semi-analytical solution with the recorded gas production data.Results indicate that the radial permeability of the shale determined using methane is in the range of 1 × 10 6-1 × l0-5 mD and decreases with a decrease in average pore pressure.This is contrary to the observed change in permeability estimated using helium.Bedding geometry has a significant influence on shale permeability with permeability in parallel bedding orientation larger than that in perpendicular bedding orientation.The superiority of the VPG method is confirmed by comparing permeability test results obtained from both VPG and SPG methods.Although several assumptions are used,the results obtained from the VPG method with reservoir gas are much closer to reality and may be directly used for actual gas production evaluation and prediction,through accommodating realistic pressure dependent impacts.