Molecular simulations were used to examine noble gas separation performances of MOF adsorbents and membranes. Grand canonical Monte Carlo simulations were combined with equilibrium molecular dynamics to compute adsorption and diffusion of Xe/Kr, Xe/Ar and Xe/Rn mixtures in 115 different MOF5. Several adsorbent evaluation metrics such as selectivity, working capacity, sorbent selection parameter, per cent regenerability were computed for each gas separation to identify the most promising MOFs. Relations between adsorption selectivity and structural properties of MOFs were also investigated to provide structure-property relationships that can serve as a guide for future experimental studies to design better adsorbents. Materials with pore sizes of 4.3-6.8 angstrom, surface areas of 150-1000 m(2)/g and porosities of 0.37-0.58 were found to be the best adsorbent candidates for Xe/Kr, Xe/Ar and Xe/Rn separations. Molecular simulations were then used to model MOFs as membranes for these gas separations. Membrane selectivities and gas permeabilities of 115 different MOFs were computed and a large number of MOFs was identified to outperform traditional polymer and zeolite membranes. MOFs with pore limiting diameters in the range of 3.9-5.7 angstrom were found to be the most promising membrane materials with high selectivities and high gas permeabilities. Our results showed that MOFs have the potential to replace traditional adsorbent and membrane materials in noble gas separation processes. (C) 2017 Elsevier Ltd. All rights reserved.