The landscape data used was interpreted from the Land MSS and TM satellite imagary for the middle and upper reaches of Dongting Lake watershed in 1980, 1995, and 2000. With the Maximum Likelihood method, noise, concealing trends which the hydrological indices represented, was eliminated. Hydrological indices include the yearly peak stage, the peak discharge, the yearly amount of runoff, and the yearly amount of sediment, values for which came from the hydrological stations located in the lake inlet of Xiang River, Zi River, Yuan River, and Li River. After being treated by the Panel Data Model, these landscape and hydrological data were analyzed by the method of Gray Connection analysis and the principal component analysis. The results showed: change of the peak stage is affected mostly by the landscape pattern change, and yearly sediment amount has impacted little affect. The peak stage change, lay on the increase or reduction of the landscape types, such as the mountain paddy fields, reservoirs and ponds, and shrub land and open forest land, whose influence values were -48.5, -48.1, and 45.2, respectively. The yearly peak discharge change is affected by the change of landscape types, including reservoirs, ponds, paddy fields (except mountain paddy fields), and shrub land and open forest land, whose influence values are -41.9, 41.2, and 41.2, respectively. Because the structure of the forest ages is not clear and the kinds of trees are simple, the effect of forest land on the change of hydrological characteristic indices is small, and the influence values are only -10.1, -13.5, and -14.5, respectively.