Abstract: Canopy reflectance of winter wheat infected with different levels (namely different Disease Incidence) stripe rust were measured through artificial inoculation. The DI (Disease Incidence) was also investigated at the same time. Further, the correlation between the DI and the spectral data was analyzed. 630～687 nm, 740～890 nm and 976～1350 nm were found as sensitive bands to DI of winter wheat stripe rust. Some inversion models for monitoring winter wheat stripe rust, therefore, have been established. The spectrum feature and difference of winter wheat stripe rust at green, yellow and near infrared band were studied qualitatively and quantitatively. The canopy reflectance at green, yellow and near infrared bands was down, up and down, respectively and the chlorophyll content was down when the stripe rust disease becomes more serious. And the red edge position shifted to the blue. It indicated that hyperspectral has the ability to diagnose the early disease symptom within the best forecast time (under 5% of total leaves infected by stripe rust) and provided theoretic foundation to further monitor winter wheat stripe rust at large scale using airborne and airspace remote sensing data.