Abstract:A case study with Eucommia ulmoides plantation in the mountainous region of North-China, predictability and fractal characteristics of temporal variation of trees transpiration were analyzed by using the method of reconstructing phase space when the lag time is 30 minutes. Transpiration measured with the technique of thermal dissipation probes in 2005 and 2006, The results showed that temporal variation of the plantation transpiration is a chaotic motion and characterized with fractal pattern and predictability during the main growth season. The fractal dimension (D), Renyi entropy (K2), embedding dimensions and the average predictability time scale of time series for the trees transpiration were 1.673, 0.002 49, 9 and 401.0 minutes in 2005,respectively, and 1.525, 0.002 53, 10 and 394.8 minutes in 2006. There existed significant difference in external appearance of temporal variation or in the amount of transpiration between the two years, but no difference about D and K2 at the same lag time. It is concluded that there were no differences in natural characteristics or internal patterns of time series in different years.