While polymeric membranes are conventional for gas separation processes, significant improvements remain possible and thereby the search for novel polymers is still on-going. The present study provides a way to develop structure-property relationship for polyimides and copolyimides, in order to lead new experimental studies with respect to recommendations on tunable monomers with promising transport properties for specific applications. This method advances the group contribution study based on molar volume contributions of subunits proposed by Robeson et al.  for the prediction of He, H-2, O-2, N-2, CO2 and CH4 permeability parameters and O-2/N-2, CO2/CH4, H-2/CO2, H-2/CH4, CO2/N-2, and He/N-2 perm-selectivities of 490 polyimide and copolyimide structures. The database is screened to identify the high-performing subunits among the 107 considered, then the defined permeability contributions and volume ratios of such subunits are used to predict the gas separation performance (i.e., selectivity and permeability) of the resulting copolyimides. Firstly, the results indicate enhanced agreement between experimental and predicted transport properties, presumably due to the division of the polymer structure into large subunits, the definition of volume quantities of subunits, the expanded database and the consideration of a specific polymer class (namely, polyimides and copolyimides). Secondly, the CO2/N-2 and He/N-2 gas separation properties cannot be further improved beyond the selectivity-permeability trade-off bound even with a judicious coupling of high-performing subunits, but separations of other gas pairs can exceed the existing upper bound by the incorporation of subunits with sulfone or side methyl groups, used as precursors of thermally rearranged polymers, or with spiro-centered or bridged bicyclic features. One of the key contributions of this study is the recommendation for the synthesis of the polymers represented by the predicted data points above the trade-off bounds to further enhance gas separations.