Abstract: Abstract: Leaf potential fluctuation comes from the changes of membrane potential in leaf cells, it caused by ion transporting across cell membranes and related to electrical coupling between cells. Understanding the life information behind leaf potential fluctuation is of great significance in study plant signal transduction, stress resistance evaluation, ecological and environmental monitoring, growth regulation, precision agriculture and many other fields. However, the leaf potential fluctuation of plant shows a very complex and non-stationary property. This complexity of the leaf potential fluctuation is the characteristic of life activities, but it brings great difficulties to analyze the information from the leaf potential fluctuation. Traditional signal analysis methods based Fourier Transform are only suitable for linear non-stationary signal processing. Different signal length leads to different results. Wavelet transform needs to pre-selected wavelet basis function, different wavelet basis will produce different results. Hilbert-Huang transform used in this paper is a new signal analysis method, this method avoids the defects of traditional methods that using stationary signal to compose non-stationary signals. Since Hilbert-Huang transform has higher resolution in both frequency domain and time domain, it is a more adaptive time-frequency localization analysis method. Thus, the interpretation about the potential fluctuation of plant leaves based on Hilbert-Huang transform can be more accurate. In this paper, the maize seedlings were treated by polyethylene glycol (PEG) solution of -0.1MPa to form osmotic stress, the leaf potential fluctuation in maize leaves was acquired after osmotic stress 0, 1, 2,3 and 4 days, the Hilbert-Huang transform was used to analyze the variation rule of the leaf potential fluctuation. After made the Hilbert-Huang transform of the leaf potential fluctuation signal of maize seedlings under different stress days, respectively, Hilbert spectrum and marginal spectrum of the leaf potential fluctuation signal was obtained. The marginal spectrum characteristic parameters such as spectral edge frequency (SEF), spectral center frequency (SCF), marginal spectrum entropy (MSE) and action potential sensitive index (Q) were calculated. While acquiring the leaf potential fluctuation signal, malondialdehyde (MDA) content and chlorophyll content in maize leaves under osmotic stress were also measured and analyzed. The results showed that the marginal spectrum of the leaf potential fluctuation on maize seedling leaves was continuous spectrum which frequency distributed in 0.5Hz or less. The SEF and SCF of the marginal spectrum showed a trend of increase first, thenit decreased and increased again along with osmotic stress days. The trend of action potential sensitive index Q was opposite to that of the SEF and SCF. The study also found that the changes of the MSE about the leaf potential fluctuation on maize seedling leaves was increased first and decreased afterwards with stress time. By comparing the changes of the parameters about marginal spectrum of the leaf potential fluctuation on maize seedling leaves during osmotic stress and changes of physiological indices MDA and chlorophyll content, we found that the MSE peak time could be used as sign of self-regulation and adaptive responses limits of leaf cells under osmotic stress, the Q value could be used as a sensitivity standard of maize leaf cells responsiveness to osmotic stress. According to the changes of the marginal spectrum characteristic parameters of leaves potential fluctuations in maize seedling under osmotic stress, it was possible to realize real time, in-situ and nondestructive evaluation (NDE) of maize seedling leaf cells functional status.