The objective of the research is to apply hyperspectral technique into eco-restoring monitoring. Through the ASD Fields HH portable field spectrometer, the hyperspectral data of dominant plant species in vegetation at different eco-restoring stages in semi-arid grassland in Helin County, Inner Mongolia were collected. The original spectrum reflected data were pretreated by wavelet threshold denoising through ViewSpecPro software before analysis. Using the first derivative spectra between 660 and 800 nm, and the methods of detrended canonical correspondence analysis (DCCA) by Canoco 4. 5 software, the canopy hyperspectral datum of 6 dominant plant species was calculated. The results indicated that the dominant plant species at early succession stage were Setaria viridis and Caragana microphylia, at 5 years eco-restoring stage they were Salsola collina and Caragana microphylia and at late succession stage they were Pinus sylvestnis var. mongolica and Salsola collina, same as field survey. The graph of DCCA indicated that the influential bands of dominant species canopy at early eco-succession stage were short bands, with a large variation among species, the influential bands at 5 years eco-restoring stage were near infrared bands between 1 000 and 1 050 nm, and that at late stage were near infrared bands of 1 040-1 075 nm. The DCCA also showed obviously differences in canopy spectrum among 6 dominant species, and obviously differences among 3 eco-restoring stages.