A few abnormal data stored in harmonic current monitoring data affects the accuracy of the harmonic source localization and harmonic division of responsibilities. For this problem, according to the statistical characteristics of the measured harmonic current data, the harmonic current monitoring data distribution model is established using three-parameter Weibull probability distribution, the least squares method is used to estimate the shape parameter, scale parameters and location parameters of Weibull distribution. On the basis of the three-parameter Weibull distribution parameter estimation results, the total sample interval is obtained in order to determine the harmonic current anomaly threshold. The simulation results show that Weibull distribution has better adaptability for harmonic current characteristic, the method proposed in this paper can identify the abnormal data effectively. This work is supported by National Natural Science Foundation of China (No. 51407150).