Abstract: Precipitation is the most critical and important in grid element prediction. The prediction performance of precipitation directly reflects the prediction ability and level of grid elements. Firstly, various test methods for precipitation in numerical weather forecast are reviewed, mainly including: classical test based on dichotomy contingency table, spatial test based on attribute and scale characteristics, ensemble forecast verification and verification for extremely rare events. Secondly, the post-processing techniques of precipitation statistics are summarized: first, the direct statistical post-processing of precipitation products based on model output and the indirect post-processing of objective diagnosis based on various elements of model output; Second, for parameterized and nonparametric precipitation of ensemble prediction and post-processing correction method associated with the structure of prediction variables; The third is the exploration and attempt of meteorologists on precipitation correction based on artificial intelligence and big data technology. Finally, the advantages and disadvantages, objectivity, applicability and problems needing attention in the operation of various test methods are discussed, and the revised methods of grid precipitation postprocessing are summarized and discussed.