Abstract: A multi-scale information entropy (EEMD-ME) method based on ensemble empirical mode decomposition (EEMD) was proposed to quantify the stochastic characteristics of monthly precipitation across the different time scales during 1961-2016 at 138 meteorological stations  in the Yangtze River Basin. Then, the spatial categorization of meteorological stations was performed using the fuzzy C-means clustering (FCM) algorithm in the basin. Finally, the lag-time correlation between average monthly precipitation series in each sub-region and NINO1+2 index was discussed. The results are as follows: (1) The monthly precipitation in the Yangtze River Basin exhibits remarkable seasonal, interannual and interdecadal variation characteristics. (2) The randomness of monthly precipitation in the east of the basin is higher than that in the west under the different time scales. The randomness of IMF component of monthly precipitation decreases with the increase of time scale, and the difference of randomness among different regions is more and more obvious. (3) The stochastic characteristics of each IMF component of monthly precipitation increases gradually from west to east along the latitudinal direction, and shows quasi uniformity along the longitudinal direction. (4) The monthly precipitation in the basin can be divided into six homogeneous sub-regions: the western plateau region, the Hengduan mountain region, the northern low mountains and basin region, the southern low mountains and hills region, the southeast Poyang lake plain region and the eastern Yangtze River Delta region. (5) The optimal time lags of average areal monthly precipitation responding to NINO1+2 are diverse across the different sub-regions, and the lag period increases gradually from 2 months to 4 months from the southeast coast to the inland area.