Abstract: To realize effective monitoring and early warning of Potato virus Y(PVY)infection via spectral analysis, leaf spectra from PVY-infected tobacco with different severities were obtained by using ASD non-imaging spectrometer. The analysis results showed that with the increase of PVY severity, the leaf spectral reflectance increased gradually in the visible band(400-700 nm)and decreased gradually in the near-infrared band(700-1 300 nm). The two bands significantly correlated with the severity of PVY-infected tobacco were screened by unitary linear recursive analysis. Based on the spectral reflectance at the band combinations of 453, 518, 572, 708 and 775 nm, a stepwise regression diagnosis model for PVY severity grade was established. The verification decision coefficient (R²) of the model was 0.67 and its root mean square error(RMSE)was 1.97 with the prediction accuracy of 83.84%. Therefore, the spectral reflectance of tobacco with different severities of PVY infection was significantly different in specific bands. The prediction model established by the stepwise regression method could be used to diagnose PVY severity.