Abstract: Multi-source data is a complex data type that integrates multiple information sources and its main feature is that different information sources describe the samples and the relationship between the samples from different perspectives.Specifically in the field of distribution networks, the data obtained by different measurement systems for the same node is different,and even there are large differences. This paper presents a comprehensive utilization method of multi-source data suitable for the calculation of distribution network line losses. Firstly, to facilitate data fusion, multi-source measurement data are converted. Secondly,to ensure the time consistency of the measurement section,a certain measurement time of PMU is selected as the benchmark,time registration and data filling operation are carried out on SCADA system data,and the smart meter data is adopted after the time scale is aligned with the measured value and predicted value,and the time series data is filtered to obtain more accurate distribution network data. Furthermore,the multi-source data fusion is realized based on D-S（Dempster-Shafter,D-S）evidence theory method. The line loss calculation based on the high-quality data after fusion can make more refined use of the distribution network data,and make up for the lack of information and low accuracy caused by the extensive data processing of the traditional method. Finally,a 10 k V distribution network is taken as an example to analyze and calculate the distribution network line loss results,and the result shows that the proposed method can better complete the comprehensive utilization of multi-source data and improve the accuracy of the distribution network line loss calculation.