The aim of this paper is to design a new normalized difference vegetation index, which is insensitive to pigment contents and relative water contents, but is sensitive to leaf area index (LAI). The canopy reflectance of winter wheat which was infected with yellow rust of different severities, and the LAI, canopy chlorophyll density (CCD) and relative water contents (RWC) of the whole wheat were measured respectively. Through regression analysis and test, the result indicated NDVI, (SDr − SDb)/(SDr + SDb) and (R900 − R1450� /(R900 + R1450� had high precision for estimating LAI. The sensitive relationship had been analyzed between above three indices and CCD and RWC, respectively, the result showed the near-infrared (NIR) and short-wave-infrared (SWIR) normalized difference index (R900 − R1450� /(R900 + R1450� was most insensitive to CCD and RWC. Moreover, the index was not easiest to get to the saturation level than the other two indices. Therefore, the index (R900 − R1450� /(R900 + R1450� is a relatively good index for the estimation of wheat LAI, and the model determination coefficient (R 2 � is 0.7924, and RMSE is 0.59, and relative error is 23.6%. This study not only provides a new method for the estimation of LAI by using hyperspectral remote sensing, but also offers technology and information support for precision agriculture.