A mean generating function algorithm is used to interpolate the missing precipitation observations over eastern China during the period of 1900－1950. Compared with the grid precipitation of Climate Research Unit (CRU) and the reconstructed precipitation of Wang et al.(2000), the time series (1900－2006) of monthly rainfall indices in North China, the middle-lower reaches of the Yangtze River, and South China of eastern China are constructed. The scanning t-test detection algorithm is used to analyze and detect multi-scale regime shifts in three time series of regional precipitation. Results indicate that during the last 100 years the precipitation over the middle-lower reaches of the Yangtze River has experienced five interdecadal regime shifts, whereas the precipitation over South China and North China has very likely undergone four interdecadal regime shifts. Furthermore the regime shifts of regional precipitation occurred simultaneously in the early 1920s, the middle 1940s, the middle 1960s and the middle 1970s, respectively. That is to say, the precipitation over the middle-lower reaches of the Yangtze River and South China experienced an interdecadal decrease in the early 1920s. Meantime precipitation over North China underwent an interdecadal increase. The precipitation in the three regions experienced an abrupt increase synchronously in the middle 1940s. In the middle 1960s, the precipitation in North China decreased, on the contrary, the precipitation over South China and the middle-lower reaches of the Yangtze River experienced an interdecadal increase simultaneously. The precipitation over South China and North China experienced an interdecadal decrease in the middle 1970s, while the precipitation over the middle-lower reaches of the Yangtze River underwent an interdecadal increase in the early 1980s. It is also revealed that the time scales of regime shifts in precipitation provide a clue that should be used to predict the duration of climate regime for precipitation after the transition occurs.