With the rapid development of online social networks, such as social networking services, microblog, online community, etc., a real two-way communication and new media age has been gradually forming. Everyone can create their own content and spread out quickly through online social networks. According to incomplete statistics, Sina microblog generates over 1000 new microblogs per second and the daily incremental size of data is 5TB. Thus, massive data has brought severe challenge to public opinion monitoring. This article will introduce a microblog public opinion monitoring system based on the Hadoop. It can mine and analyze large scale collected data, realize detection and tracking of hot topics, perform social network analysis on the microblog and visualize the analysis result. The proposed system will provide automated, systematic, and scientific information support for party and government organizations, enterprises and other units and organizations to detect sensitive information timely, grasp the hot points and the trend of public opinion and deal with the crisis of public opinion.