Abstract: Based on the analysis of public meteorological service data fl ow, this paper proposes a message-driven data processing technology in the meteorological service to improve the timeliness of meteorological data services and the uniformity of data resources. The key technologies, such as message middleware, system architecture, interactive fi le standard, are introduced here. An example of PM-MQPF data is described in detail to express the advantages of data fl ow, which was driven by messages in meteorological data services system.