Methane adsorption in porous carbon systems such as coal and the organic matrix of gas shales is an important factor in determining the feasibility of CO2 injection for enhanced natural gas recovery and possible sequestration of CO2. Methane and CO2 adsorb competitively on carbon surfaces and an understanding of each gas individually is important for determining a model to predict the feasibility of this approach for permanent CO2 storage. Coal and gas shales have a very heterogeneous pore system, ranging from the micro, meso, and macro-scales, with the pore size strongly affecting the adsorption behavior. In micropores, the force fields of opposing pore walls are close enough that they will overlap and significantly influence the adsorption behavior, which affects adsorbate packing and density. To determine the size at which these effects become non-negligible and to determine the magnitude of this impact, grand canonical Monte Carlo simulations have been carried out to estimate the adsorption isotherms of methane across a range of pore sizes and at various temperature and pressure conditions characteristic of subsurface conditions. These isotherms have been calculated on graphitic surfaces as an initial model of coal and kerogen of gas shales. The general trend within pore sizes is that larger pores exhibit lower excess density compared to smaller pores. However, at pressure above I MPa, the adsorption capacities of 0.6-nm pores drop below those of the wider pores, ultimately decreasing below that of the 1.2-nm pore at 18 MPa. The density of adsorbed methane changes non-monotonically with increasing pore width, and drops to a minimum in 1.2-nm pores at 12 MPa. The isotherms have been compared with experimental data to gauge their accuracy, and the behavior of the adsorbed layer has been examined in detail. At pressures less than 2.5 MPa, the molecular simulation estimates underpredict the excess adsorption, while at pressures greater than 2.5 MPa up to 20 MPa, the simulation estimates overpredict the excess adsorption. This discrepancy is likely due to the limitation of the experimental-based model that was used to generate the pore size distribution and the surface functionalities of the porous media that were ignored in the molecular simulation investigations, but likely play an important role in determining accurate capacities under confinement at the nanoscale. (C) 2013 Elsevier B.V. All rights reserved.