Amorphous materials are usually characterized using nitrogen adsorption isotherms at 77 K taken at pressures up to 1 bar to obtain pore size distributions. Activated carbons are amorphous microporous graphitic materials containing pores which can range from nanometers to microns in width and which can, in principle, be tailored to adsorb specific molecules or classes of molecule by changing the method of preparation (the activation process). For the physical chemist, they pose the challenge of understanding how gases adsorb in graphitic nanopores, that is, in restricted geometries, and of using that understanding to improve their characterization. In this paper, we compare pore size distributions of an ultrahigh surface area activated carbon (AX21) determined from nitrogen adsorption measurements up to 0.6 bar at 77 K with those determined from carbon dioxide adsorption measurements up to 20 bar at 298 K. Our analysis employs grand canonical and Gibbs ensemble Monte Carlo simulations together with accurate site-site interaction models of the adsorbates. We find that the calculated pore size distributions for each adsorbate are quite different, and the adsorption of one gas can be estimated from the adsorption of the other gas to within an error of 25% at the highest pressures only. At lower pressures, we speculate that large errors are due to the behavior of nitrogen in carbon micropores in which diffusion is severely limited. To substantiate this speculation, we have calculated the self-diffusion coefficient for nitrogen at 77 K and carbon dioxide at 298 K in carbon slit pores using equilibrium molecular dynamics. The results suggest that nitrogen is diffusionally limited, and possibly frozen, in such pores whereas carbon dioxide remains mobile. We conclude that room-temperature carbon dioxide adsorption isotherms up to the saturation pressure could provide a more accurate characterization of carbon microstructure than nitrogen isotherms at 77 K up to 1 bar.