Abstract: Fractured?vuggy carbonate reservoir is characterized by varied models of reservoir development, complex storage spaces, serious heterogeneity, and great development difficulties. This paper presents comprehensive clustering evalation of its depletion drive characteristics, through modified fuzzy C?means (FCM) analysis, develops a new approach for comprehensive evaluation of such a reservoir by depletion drive process. Based on the well production performance data, it introduces the discrete particle swarm optimization algorithm to overcome the inherent limitations of conventional FCM clustering analysis that depends on the arbitrary setting the number of clusters and the initial clustering center, and easily getting stuck in local optimal solution. Using this approach, a total of 62 key oil wells in M fracturedvuggy carbonate reservoir are classified and evaluated comprehensively into four categories and eight subclasses, providing the classification standard of dynamic attributes of all kinds of oil wells. The case study indicates the reliability of this approach for differentiation comprehensive treatment and preparation of EOR countermeasures