Abstract: Based on daily precipitation observations from the national meteorological observatories,daily outgoing longwave radiation data from NOAA,and reanalysis data from the NCEP (National Centers for Environmental Prediction),an extended range forecasting for the low-frequency precipitation over south of the Yangtze River during April,May,and June (AMJ) from 1979 to 2013 was carried out under different background circulations.The atmospheric circulations over tropic or mid-high latitude,which are significantly correlated with the 10-30 d low-frequency precipitation,were selected as the influencing factors.The results show that the 10-30 d component of the AMJ precipitation over south of the Yangtze River is the most significant.The precipitation anomalies are significantly related to the tropical convection locates over the eastern Indian Ocean and Indonesia,as well as the atmospheric circulation anomalies over the high latitude of Europe and Siberia.They can be selected as the predictors for the extended range precipitation.When the geopotential height anomalies over the high latitude of Europe and Siberia areas appear to be suppressive,while over North America and west to the Lake Baikal appears to be active,moreover,the tropical convective appears to be suppressive,the rainfall over south of the Yangtze River turns scarce.The center of the rainfall anomaly locates in the middle and lower reaches of the Yangtze River.Based on the spatial distributions of geopotential height anomalies at the period longer than 30 days at 500 hPa,the low-frequency background of daily precipitation is divided into three categories.The correlation in -30 lag days between low-frequency precipitation and predictors exhibits their individual features in each background.The 500 hPa low-frequency circulation at the period longer than 30 days provides the stable background state for the 10-30 d extended range variations.The correlation between low-frequency precipitation and predictors turns out to be more significant and the lag days of correlation becomes longer,considering different backgrounds.