Blending is a strategy to improve the permselectivity of gas separation membranes by combining the advantages of two or more polymers. The blending of polymers results in miscible or immiscible morphologies, which have distinctly different properties and performance benefits. Perfluoropolymers, specifically Teflon AF1600 and Hyflon AD60, have unique chemical properties that make them attractive as membranes for natural gas sweetening. Here, blended membranes of these two perfluoropolymers are generated, with miscible and immiscible morphologies achieved by varying the evaporation rate of the solvent. This ability to achieve two different blended morphologies is associated with the different solvability of the two perfluoropolymers in the solvent, not the miscibility of the polymers with each other. The miscible blended membranes had higher density than the pure polymers and CO2 sorption is bounded by that observed for pure Teflon AF1600 and Hyflon AD60 membranes, while CH4 sorption is depressed in the miscible blended membranes. Correspondingly, the CO2/CH4 selectivity of the miscible blended membranes was improved relative to the two pure polymer membranes. In contrast, the immiscible blended membranes had lower density and a more open morphology, especially the Teflon AF1600 discrete domains. This increased CO2 and CH4 permeability, compared to the pure polymer membranes, with comparable CO2/CH4 selectivity. To model the resulting performance, blended membrane mixing theories were applied; standard approximations based on the permeability of the pure polymer membranes and more rigorous modelling based on dual-sorption theory of the blended membranes sorption and diffusivity. The standard approximation models were unable to correlate with the experimental data for these blended membranes, while the dual-sorption based model was significantly more accurate. Therefore, to predict permselectivity of blended membranes performance it is more appropriate to employ the rigorous dual-sorption model.