The invention provides a line-loss anomaly detection method for an electric power measurement automation system. The method comprises the steps of S1, acquiring a line loss daily fluctuation index and a line loss variation tendency index of each line through the calculation of a line loss data table; S2, respectively judging whether the line loss daily fluctuation index of each line exceeds a first threshold value or not and respectively judging whether the line loss variation tendency index of each line exceeds a second threshold value or not; S3, judging a line with the line loss daily fluctuation index thereof to exceed the first threshold value or the line loss variation tendency index thereof to exceed the second threshold value as a line loss abnormal line. According to the technical scheme of the invention, two abnormal conditions, namely the line loss mutation anomaly and the line loss variation tendency anomaly, can be found out based on the line loss daily fluctuation index and the line loss variation tendency index of each line. Meanwhile, through adjusting the above threshold values, the anomalies of different degrees can be found out. The number of daily abnormal lines can be adopted to evaluate the health degree of the entire power grid. Therefore, the finding sensitivity of the existing system for line loss anomalies is improved. The defect that the manual inspection and discovery is not timely can be overcome. The reasons for line loss anomalies can be further detected.