Abstract: By means of combining NWP outputs and historical observation data,a precipitation probability forecast method based on relative humidity is established.Fine-grid forecasting results of relative humidity from EC model during the year 2012 and 2013 in eastern China are classified into different vertical layers and different moisture grads.The historical precipitation probability corresponding to these layers and grads are analyzed to find out a relationship between the forecast humility and precipitation probability.The relative humidity predicted by the current forecast model is then used to estimate the future precipitation probability.Different methods such as ETS score,Brier score,reliability map and case study are used to verify the forecasting effects of precipitation from July 2013 to May 2014.The results show that the method can produce fairly good distribution of future precipitation.The effects of synoptic-scale rain band appear to be better than those of convective precipitation.The false forecasting rate is effectively reduced compared with EC model results.The precipitation production can be used to specify the location and probability of the precipitation and provides a better reference value.