Firstly, the common electricity-stealing methods and its characteristics were introduced and the characteristic parameters related to electricity-stealing were researched. Secondly, the relationship among the various data, the abnormal events from electric power information acquisition system and the behavior of electricity-stealing was established. Thirdly, an Electricity Anti-Stealing structure model based on electric-power information acquisition and big data was designed, including data preprocessing, power use anomaly detection model and electricity-stealing suspected prediction model. Finally, the validity of the structured Electricity Anti-Stealing model was proved by the validation of the experimental data. This paper provides an effective method to solve the problem of electricity-stealing monitoring under big data conditions.