Abstract: This study analyzes impacts of cumulus parameterization schemes on summertime precipitation forecast in China using the WRF model. Five sets of simulations using the Kain-Fritsch, the multi-scale Kain-Fritsch, the Tiedtke, the new Tiedtke and the scale-aware new Tiedtke schemes were conducted. The horizontal resolution of the simulations is 9 km. The results show that the scale-aware schemes (multi-scale Kain-Fritsch and scale-aware new Tiedtke schemes) perform better in the forecast of the magnitude and location of precipitation as well as the diurnal variation. In regard to the probability density distribution of precipitation, compared to the Kain-Fritsch and the new Tiedtke schemes, the scale-aware schemes show lower (higher) frequency of small and middle (high) level of precipitation, leading to more comparable frequency of precipitation lower than 50 mm/d with observations and overestimation of frequency of precipitation higher than 50 mm/d. Further comparisons indicate that the precipitation simulated by the Kain-Fritsch and the new Tiedtke schemes (two scale-aware schemes and the Tiedtke schemes) are dominated by convective rain (microphysical rain). The percentage of convective rain dramatically decreases to 30% in the multi-scale Kain-Fritsch and the scale aware new Tiedtke schemes, and only accounts for 15% of total precipitation for precipitation higher than 50 mm/d. Statistical results show that the scale-aware schemes help to reduce the overestimation of light rain and the underestimation of heavy rain. The threat scores of 24 h accumulated precipitation are higher in the simulations using scale-aware schemes than those using the original schemes for precipitation within the range 0.1 mm to 25 mm.