The invention provides a network traffic base line self-learning adaptive method, comprising the steps of collecting and storing network traffic data, calculating network traffic base line values, determining a dynamic base line critical value and updating the base line values and the critical value. According to the technical scheme provided by the invention, when a service traffic fixed threshold is not set, important traffic anomaly alarm information can be provided for a maintainer, a network manager is effectively assisted in sensing and discovering an anomaly in the network as soon as possible, and a network traffic data trend can be visually reflected.