This work aims to extend previous studies about performance of gas separation through the carbon molecular sieve membrane (CMSMs) in different conditions by employing statistical analysis and modeling to find the optimal pyrolysis and operating conditions. General D-optimal design is applied to optimize gas permeability and selectivity by implementing five main parameters consisting of precursor materials, blend composition, final pyrolysis temperature, vacuum pressure and operating pressure. Results from statistical analysis showed that each of these five variables plays a significant role in performance of carbon membranes. Also findings showed that pyrolysis temperature was the dominant factor among the others; whereas operating pressure was almost a neutral one. The optimal condition is the blend composition of 75% Matrimid, pyrolysis temperature at 745.1 degrees C and vacuum pressure of 1.0E-07Torr. Under these conditions, the model estimated a CO2/CH4 selectivity of 133.7. These developed models can be employed as a useful technique for gas transport optimization.