Mesoscale weather phenomena have a significant impact on human production and life. Mesoscale numerical prediction models are one of the main tools used for numerical weather prediction. Owing to the limitations of model resolution and insufficient understanding of the physical mechanisms of weather phenomena, many complex weather processes can only be implicitly expressed by parameterization schemes, the research of which is helpful in promoting the continuous optimization of the simulation and prediction effects of mesoscale numerical prediction models. Based on the characteristics of typical mesoscale numerical prediction models at home and abroad, the research status of cumulus convective, cloud microphysical, planetary boundary layer, land surface process, and radiation transfer process parameterization schemes, which are important according to the main forms of atmospheric motion, have been summarized. The theoretical basis and application scenarios of representative parameterization schemes are also compared in the present study. The main research paradigms in this field are comparative experiments that include the comparison of different scheme effects in simulating the same weather phenomenon, comparison of the same scheme effects in different weather scenarios, and optimization experiments on independent or combined schemes. In the future, the physical mechanisms of various weather phenomena and their influencing factors will be explored in depth, and parameterization schemes will be coupled at the levels of multiple elements, scales, and schemes. Subsequently, the simulation of the gray area will receive more attention, and the application options of parameterization schemes will be more diversified. Owing to the arrival of the big data era and the high demand for data analysis, parameterization schemes and machine learning will be developed to form a new mechanism driven by methods and data.