Abstract. In this study, a novel approachser-induced breakdown spectroscopy (LIBS) is proposed to rapidly diagnose stem rot (SSR) in oilseed rape ( L.). A rapid diagnostic method is important to prevent this worldwide disease and promote growth of oilseed rape. A total of 120 fresh leaves, including 60 healthy and 60 SSR-infected leaves, were collected to acquire LIBS spectra. Robust baseline estimation (RBE) and wavelet transform (WT) were applied to preprocess the raw LIBS spectra for baseline correction and denoising. K-nearest neighbor (KNN), radial basis function neural network (RBFNN), random forest (RF), and extreme learning machine (ELM) methods combining full LIBS spectra were chosen to establish classification models to identify healthy and SSR-infected leaves, and the ELM model obtained classified accuracy of more than 80.00% in the prediction set. Twenty-four emission lines were selected by second-derivative spectra as the most relevant to distinguish healthy and SSR-infected leaves. The ELM model using the optimal emission lines improved the classified accuracy to more than 85% and the specificity to 95.00%. Compared with full-spectra models, the number of variables in the models based on optimal wavelengths was reduced from 22,036 to 24, a reduction of 99.89%. This study indicates that LIBS combined with appropriate chemometric m. Keywords: Chemometrics, Laser-induced breakdown spectroscopy, Oilseed rape, Sclerotinia stem rot.