We discuss the design of optimum graphitic microstructures for selective adsorption from gaseous mixtures. Binary mixtures of methane and nitrogen at 293.1 K and pressures up to 1 bar were chosen as model systems. Non-local density functional theory (DFT) was used to predict the adsorption of nitrogen and methane in slit-shaped pores for a pore size range of 0.5-5 nm. Predictions of gas mixture adsorption and selectivity factors were obtained from ideal adsorbed solution (IAS) theory. From the selectivity as a function of pore size and pressure it is possible to identify optimum structures for methane/nitrogen separations under a given set of conditions. Using our data we predict selectivities for the high surface area carbon AX21.