Precipitation is a discontinuous observation element, and the accuracy of precipitation observation data cannot be well checked by conventional quality control methods, so we carried out the study on quality control technology of precipitation observation data from automatic weather stations (AWSs). In the study, a method for quality control of precipitation data of AWSs was proposed based on multi-source data fusion such as the data of radar and satellite. A processing algorithm for radar quantitative precipitation estimation products based on standard Z-R relationship was studied, and the one-hour radar quantitative precipitation estimation of AWSs observations in Ningxia was generated. The technical process of quality control of precipitation data of AWSs based on the radar and satellite precipitation estimation products was established. Moreover, the quality control of precipitation data of AWSs was conducted, and the results were evaluated. The results show that: (1) For national stations, the quality control method of AWSs precipitation data based on multi-source data fusion and the quality control of precipitation data using MDOS are both "correct" with a higher consistency rate, reaching 94.88% on average. For regional stations, the consistency rate of "correct" is significantly lower than that of national stations, 51.57% on average. (2) The quality control method of AWSs precipitation data based on multi-source data fusion produces more "suspicious" and "wrong" data than MDOS, and the detection rate of "suspicious" and "wrong" data is higher. (3) The quality control method of AWSs precipitation data based on multi-source data fusion is more realistic than MDOS quality control for the quality control of erroneous precipitation data. Therefore, the quality control method of AWSs precipitation data with multi-source data fusion can better solve the quality problem of AWSs precipitation data in our district, and provide technical and theoretical support for the quality control of precipitation data in our district.