In this study, the challenge of selecting metal-organic frameworks (MOFs) as filler particles in high-performance mixed-matrix membranes (MMMs) was examined using atomistic and continuum modeling. We tested several theoretical permeation models, including the Maxwell, modified Maxwell; Bruggeman, Lewis-Nielson, Pal, Felske, and modified Felske models, by comparing the predictions of these models with the experimental data for IRMOF-1/Matrimid and CuBTC/Matrimid MMMs. After identifying the model making the best predictions, we examined the performance of 80 new MOF-based MMMs composed of 10 different MOFs and eight different polymers for CO2/CH4 separations. Our results show that selecting the appropriate MOF as filler particles in polymers can result in MMMs with extraordinarily high CO2 selectivities and CO2 permeabilities relative to those of pure polymer membranes. The methods introduced in this study will create many opportunities for selecting MOF/polymer combinations for MMMs with useful properties for CO2 Separation applications.