While selective adsorption of carbon dioxide has recently attracted special interest due to concerns over greenhouse emissions, separation of methane from its mixtures with carbon dioxide and nitrogen is a key operation for biogas and natural gas enrichment. Herein we present a detailed study of the multicomponent potential theory of adsorption's (MPTA) capabilities to predict published experimental adsorption data of binary and ternary mixtures of carbon dioxide, methane and nitrogen from each gas' pure component adsorption data on activated carbons Norit R1 and Calgon F-400 and on the metal-organic frameworks (MOFs) MOF-508b and MOF-5. It is observed that at temperatures below its critical temperature (304 K), and pressures in the vicinity of its critical pressure (7.37 MPa), carbon dioxide content in the binary mixtures adsorbed on Norit R1 has significant influence on the accuracy of MPTA's predictions, as sharp mixture density increases due to possible carbon dioxide phase changes cause convergence difficulties in the model's numerical iterations. This phenomenon is not observed in the studied ternary mixtures' case. Here, discrepancies are rather attributed to the inherent limitations of using four adjustable parameters to describe multicomponent adsorption at high pressure, and to the additive errors associated with the volumetric experimental measurement technique. Carbon dioxide's high selectivity due to its strong quadrupole moment is well predicted by the MPTA, on activated carbon as well as on MOFs. (C) 2014 Elsevier B.V. All rights reserved.