Abstract: In order to solve the problems of low precision, heavy workload and low efficiency in the traditional sampling method of the smart meter error monitoring, a new method for error estimation of electricity meters is proposed. First of all, the collected daily electricity data of smart meters in the distribution platform area are screened to identify the light (empty) load data. Then, the improved dynamic line loss estimation algorithm is used to judge whether there is a sudden out-of-tolerance meter in the platform area. Next, the iterative algorithm is used to reduce the amount of suspicious meters. Finally, the clustering algorithm and the correlation coefficient are used to locate the sudden out-of-tolerance meter, setting the daily average error electric quantity excess threshold to locate the slowly out-of-tolerance electric meter. The effectiveness of the proposed method in the detections of slow and sudden out-of-tolerance meters is verified by the meter data of two platform areas with different scales. Compared with the algorithm of limited memory least square method and dynamic line loss combined with the FMRLS, the proposed method has a higher error estimation accuracy with only 10.6% false detection rate. Good error estimation results can be obtained by relying on one daily electric quantity data.