The main objectives of this research were to select the best hyperspectral narrow bands in the study of rice under different levels of nitrogen and water was used to study the nutritional status of rice. The methodologies employed used partial least squares (PLS) analysis method and spectral bands inter-correlation method (ICM), on the basis of an experiment using reflectance spectra of rice leaves and the concentration of three foliar biochemicals: nitrogen, chlorophyll-a and water, we select the most appropriate wavelengths. We established several broad-band and narrow-band (wavelengths) combinations and compare the inverse effects with each other to determine the inversion accuracy of narrow bands. From the PLS regression result, we confirmed that a 5 narrow bands combination includes 552 nm, 660 nm, 675 nm, 752 nm, 776 nm is the best inversion model of rice leaf nitrogen content, and the 3 narrow bands combination includes 1158 nm, 1378 nm, 1955 nm is much available for rice leaf water content inversion.