This paper found rainfall and runoff was in a state of chaos in Poincare section and developed a method for constructing a relationship model of rainfall-runoff based on time series analysis technology. It aimed at estimating and discussing the relationship of rainfall-runoff time series by Euclidean distance (n>2). A linear model was well built to illustrate time-correlation characteristics with the correlation coefficient 0.994, which was able to predict each other by obtaining one rainfall or runoff value. Analytical results verified the accuracy of this model. It is complementary research to our previous work which discussed the special situation (n=2), and the previous work can’t be applied to predicting rainfall or runoff. So it was an in-depth study and got improved.