Hyperspectral imaging technology was applied to develop a rapid, accurate and non-destructive detection method for honeysuckle mildew degree levels. The original spectral data were analyzed by three pretreatment methods with Savitzky-Golay (SG) convolution smoothing, Multiple Scatter Correct (MSC) and SG-MSC. A comparison was made among SG, MSC and SG-MSC based on Partial Least Squares (PLS), of which the best pretreatment method was SG-MSC. The Successive Projection Algorithm (SPA) and Competitive Adaptive Reweighted Sampling (CARS) were used to extract the characteristic wavelengths after SG-MSC pretreatment. Partial Least Square Discriminant Analysis (PLS-DA) and Last Squares Support Vector Machine (LS-SVM) were applied to build discriminant analysis models based on characteristic wavelengths. The results showed that the LS-SVM model based on CARS performed the optimal discriminant performance for honeysuckle’s mildew degree levels, with the accuracy of 100% for training set and validation set. Therefore, hyperspectral imaging technology can be used to identify mildew degree in honeysuckle effectively and non-destructively based on characteristic wavelengths.