Time series of the dryness-wetness (DW) index of 531 yr (AD 1470-2000) at 42 stations in regions A (most of North China and the east of Northwest China) and B (the Yangtze-Huaihe River valley) in China are applied to investigating the historical DW characteristics over various periods of the series with a relatively stationary average value using Bernaola-Galvan (BG) algorithm. The results indicate that region A/B underwent three drought-intensive periods (DIP; 1471-1560, 1571-1640, and 1920-2000/1501-1540, 1631-1690, and 1911-1960) in the last 531 years. In the DIP of the last 130 years, the frequency of DW transition has increased in region A, but not obviously changed in region B in comparison with the other two historical DIPs. The dry period started in about 1920 in region A with severe drought events occurring from the late 1970s to the early 1980s. It lasted for about 50-70 yr in this century, and then a DW shift took place. The wet period in region B might maintain for the coming several decades. The variations of DW in region A are positively correlated with changes in temperature, but in region B, the correlation with temperature is weaker. It is found that the number of DW indices of various categories within a running window is an exponential function of the running window length. The dryness scale factor (DSF) is defined as the reciprocal of the characteristic value of the exponential distribution, and it has a band-like fluctuation distribution that is good for the detection of extreme drought (flood) clustering events. The results show that frequencies of the severe large-scale drought events that concurrently occurred in regions A and B were high in the late 12th century, the early 13th century, the early 17th century, and the late 20th century. This provides evidence for the existence of the time-clustering phenomena of droughts (floods).