Carbon molecular sieve membranes (CMSMs) have drawn substantial research attention in recent years due to their ability to overcome the trade-off limitation between permeability and selectivity that is commonly observed in polymeric membranes used for gas separation. The performance of CMSMs is governed by various factors, such as the choice of polymeric precursors and their pre-treatment, pyrolysis, and post-treatment conditions. This review examines the critical aspects in the process of developing CMSMs based on polyimide precursors for gas separation. In addition, the mass transfer mechanism and characterization methods of CMSMs are discussed. Then, the performances of various CMSMs developed so far are examined against the Robeson upper bound limit, and pilot-scale applications and an economic analysis of CMSM-based gas separation are provided. Finally, the challenges and perspective are presented as the concluding remarks.