Fusarium is a DON producing filamentous fungi which commonly infects small grain cereals. Near-Infrared Hyperspectral Imaging (HSI-NIR) is considered for its potential to manage this contamination, as it uses spatial recognition, which may be able to deal with the heterogeneity inside the batches for cereal sorting implementation. The focus of this study was the application of HSI-NIR for Fusarium Damaged Kernels (FDK) detection and DON prediction and discrimination of wheat kernels over EU limits. After the HSI scanning of 300 individual grains, the reference values were obtained attributing categories for typical fungal symptoms and analyzing DON from individual grains by HPLC. Several spectral preprocessing methods selected valuable information before model calibration. Externally validated PLS predictions showed RMSEP of 6.65 mg/kg, an R2 of 0.88 and an RPD of 3.21. However, the classification models managed wheat contaminations more appropriately, obtaining discrimination accuracies of 85.8% and 76.9% for fungal symptoms and DON at the EU limit, respectively. These findings suggest that HSI-NIR can be a suitable tool to sort DON contaminated kernels at EU limits.