Among the automatic monitoring station data, compared with the temperature, pressure and other data, quality control of precipitation data may be more complex, so it is very concerned about the precipitation data quality control in meteorological work. This paper analyzes the causes for abnormal rainfall data by minutes, by hours and the cumulative data at the meteorological automatic station. The concrete method and discrimination process of quality control is put forward. This process is used for limit inspection of real-time data, then the fine discrimination between validity and error is carried out by space, time and element match consistencies, finally, when comprehensive making decision, discriminating and processing are finished, based on human-machine combination, the results are recorded in secondary data base. Through the statistical analysis, combining the correlation between the precipitation and visibility elements of traffic meteorology, the discriminant algorithm and related threshold of quality control of the precipitation data is set up. The above quality control not only allows forecaster to avoid mistakes, but also provides the basis for timely maintenance and repair for the equipment.