ABSTRACT This study aimed to evaluate the applicability and efficiency of Near-Infrared Spectroscopy (NIR) by using dispersive NIR and Fourier Transform NIR to analyse 267 samples of Brazilian wheat flour contaminated with deoxynivalenol (DON). For this, Partial Least-squares Discriminant Analysis (PLS-DA) and Principal Component Analysis-Linear Discriminant Analysis (PC-LDA) were used as discriminatory methods. Next, the samples were classified according to the maximum tolerated limits (MTL) for DON in Brazil, 750 μg kg−1, and two groups were established for the calibration set: category A (≤450 μg kg−1), non-contaminated or below the MTL; and category B (>450 μg kg−1), contaminated or above the MTL. Validation samples through PLS-DA showed correct classification rates in the range of 85–87.5% and presented a 10–15% error; for PC-LDA, the hit rate was over 85% with an error of 10–15%. The present findings demonstrate that NIR is an excellent alternative method to classify wheat flour samples according to DON content.