On the basis of the hindcast data of the National Climate Centre secondgeneration monthly Dynamic Extended Range Forecast operational system (DERF2.0) from 1983 to 2013 from 19 observed stations in Ningxia, the predicted precipitation is evaluated by operational methods in shortterm climate prediction. The results show: the monthly averaged PS (Propensity Score) of DEFR2.0 is 53.2%, and there is little difference among the results of 1 to 10 day lead times. But the forecast quality is unstable, with PS exceeding 60% only in the April and November, and the difference between best year and worst year is nearly 30%, reflecting the large interannual variability. DERF 2.0 has better performance in the lessrain months rather than in the morerain months, as well as in spring and autumn than rather summer. There is positive correlation between the predicted pentad precipitation by DERF2.0 with the lead time of less than 30 days and actual precipitation, the greatest in the first pentad and larger from September to March next year. The predicted pentad precipitation is about 30% larger than the actual precipitation. The accuracy of predicted precipitation trends of next 1 to 6 pentads is 40.8%.The accuracy is 79.5% in the pentads with more precipitation, which is obviously higher than in the pentads with less precipitation. The accuracy of trend prediction is the smallest in January and the largest in August. The accuracy of trend prediction in the pentads with more precipitation in winter is larger than that in summer and the accuracy of trend prediction in the pentads with less precipitation in summer is larger than that in winter. The trend prediction accuracy declines with the prolongation of lead time and increases from north to south. The prediction accuracy of persistent abnormal precipitation is 60.3% by DERF 2.0 with the lead time of 4 to 14 days. The better prediction effect is found for the whole regional precipitation process. The prediction accuracy in the southern part is greater than that in the northern part for the same process, and the predicted precipitation is less than actual precipitation.