[Abstract] In order to study the optimal hyperspectral index suitable for the diagnosis of leaf water content and chlorophyll of maize plants, the relationship between different spectral wavelengths and the water content and chlorophyll content of maize leaves was analyzed through the remote sensing original spectrum and transform spectrum of hyperspectral reflectance, combined with the results of previous studies on hyperspectral index. A multi-factor model based on partial least square regression, random forest and support vector machine algorithm was constructed, and the optimal model of water content and chlorophyll content of maize leaves was screened out according to the model accuracy. The results were as follows：(1) Arbuscular mycorrhizal fungi changed the growth strategy of vegetation through symbiosis with plant roots, and promoted the increase of chlorophyll content in leaves. With the growth of plants, the difference of plant chlorophyll gradually became obvious with the difference of treatment. (2) In the model established with leaf water content and hyperspectral index, R2 of modeling set and test set ranged from 0.1876 to 0.9936 and 0.0471-0.9128, and RMSE ranged from 0.7% to 4.08% and 0.5% to 4.46%, respectively. MAE ranged from 0.007 to 0.0338 and from 0.0043 to 0.0381, respectively. In the models established by chlorophyll content and hyperspectral index of leaves, RMSE was 2.0867~11.6267 and 2.1777~7.1663, and MAE was 1.8314~8.3992 and 3.0334~10.7521, respectively. (3) The RF model was closer to the Y=x regression line in predicting leaf water content, the PLS model was more stable in predicting chlorophyll content, and the SVM model's regression curve was more closer to Y=x, providing suitable application for accurate hyperspectral remote sensing applications.