Abstract: A competency model of opinion leaders in social networks is presented, which includes three capacity factors, some explicit and implicit behavioral indicators. These characteristics of information production, information dissemination and information influence which opinion leaders should have. Based on the proposed competency model, social network users into four categories which divided: the general public, the activists, the topic opinion leaders and network opinion leaders. Then a hierarchical filtering process is designed and a system framework of filtering experimental for identifying opinion leaders is introduced. It takes the data extracted from the Sina microblog as input, then integrates some social network analysis tools such as MALLET, and develops the programs for specific functions with Python. Eventually, the opinion leaders of microblog can be identified. The experiment results prove the effectiveness of present competency model for opinion leaders and the feasibility of hierarchical filtering process.