In this paper, we present an improved method to predict the methane adsorption isotherm for a real shale sample using molecular dynamics (MD) simulation with a realistic kerogen model. We compare our simulation results both to the experiment and to the simulation results on the basis of a simple graphite model, and show how our procedure leads to the creation of more accurate adsorption isotherms of a shale sample at a wide range of pressure. A Marcellus shale sample was chosen as an example to demonstrate how to calculate the adsorption isotherms using MD simulations. Type II kerogen molecular model was selected for the dry gas window. The constructed bulk kerogen model contains mesopores (> 2 nm) and micropores (≤ 2 nm) inside. Ten different mesopore sizes of kerogen nanopore systems were constructed. According to the characteristics of methane density distribution in the simulation system, three regions can be clearly distinguished, free gas, adsorbed gas, and absorbed gas. We show that the adsorbed gas per unit pore volume increases with the pore size decreased. This is similar to previous molecular simulations with graphite model. For predicting the total adsorption isotherm of a real shale sample, both adsorbed and absorbed gas were considered. For the adsorption amount, the calculated adsorption isotherms were averaged based on pore size distribution of that Marcellus Shale sample. For nanopores smaller than 5 nm, we used total organic carbon (TOC) data to weight the absorption contribution in the kerogen bulk (i.e. inside the micropores). The total adsorption isotherm thus obtained from our simulations reproduced experiments very well. Importantly, kerogen model has overcome the difficulties of prediction using graphite models (i.e. an underestimation of adsorption under high pressure conditions) as documented in previous studies. Furthermore, we predicted the adsorption isotherms for higher temperatures. With the temperature increased, lower adsorption amount is predicted. The novelty of our improved method is that it is able to predict methane adsorption isotherm at a wide range of pressure for a shale sample by considering both adsorption in kerogen mesopores and absorption in kerogen bulk. It can be readily used for any shale sample, where the pore size distribution, porosity, and TOC are known. We remark that the above results and conclusion resulted from our simple assumption. Further discussion might be necessary.