Abstract: The ice-phase cloud microphysical process plays an important role in the forecast of cold-font snow. To study the impact of microphysical schemes on forecasts, sensitivity tests are carried out using four cloud microphysical parameterization schemes (Lin, WSM6, Thompson and WDM6) employed in the Weather Research and Forecasting (WRF) model for a typical cold-font snowstorm occurred in Xinjiang from 22 to 24 April 2014. For precipitation, the forecasts of heavy snow center in the middle Tianshan are significantly better than those in the Ili valley area, and the forecasts using the Lin scheme are the best among the four schemes. For temperature and dew point temperature, four schemes all performed well in forecasting temperature during the event and dew point temperature after the event, while over-predicting dew point temperature under 800 hPa before the event, which may be caused by the facts that the forecasting snow time was earlier and the amount of precipitation was larger than observations. In addition, the values of latent heat and sensible heat differ slightly depending on schemes while the evolutions are in quite accord with observations. From the microphysical characteristics, the evolutions of hydrometeors are considerably consistent with that of rainfall intensity, and ice-phase particles dominate. However, the amounts of hydrometeors have a large difference among the four microphysical schemes, such as graupel and snow particles are the majority in the Lin scheme, the snow and ice particles dominate in the WSM6 and WDM6 scheme, and almost all are snow particles in the Thompson scheme.