The application of traditional methods to detect feed mycotoxin is time consuming and requires a high level of experience and expertise. The objective of this study was to test possibility of using advanced molecular spectroscopic techniques to screen mycotoxin concentrations in feed grains. Barley and wheat grains are ranked as the most important feed crops for dairy cows in western Canada. However, they have been suffering from mycotoxins contamination for a long time. In this study, a total of 80 wheat and 42 barley samples were collected and detected for 6 major ergot alkaloids and 12 common mycotoxins concentration by liquid chromatography- tandem mass spectrometry. The near-IR (NIR; 680–2500 nm) and mid-IR spectra (MIR; 4000–700 cm⁻¹) of all samples were all collected with the grading NIR and FTIR. All spectra were averaged from 3 repeat NIR or FTIR measurements, each recorded from a new sub-sample. The final spectra data were imported into the UnscramblerX v10.3. Preliminary descriptive analyses were performed by both graphic tools and numerical results. To remove the spectral baseline shift, noise, and light scatter effects, 9 preprocessing methods were applied, including baseline offset, standard normal variate (SNV), detrending, SNV+ detrending, multiplicative scatter correction, first derivative, second derivative, first derivative + SNV, and second derivative + SNV. The NIR and MIR spectra were calibrated with EAs reference values using PLS technique based on different spectral preprocessing methods and selected wavelength ranges. The possibility whether we could develop fast screening methods for wheat and barley major 6 ergot alkaloids and 12 common mycotoxins detecting by NIR and MIR were revealed in this study. In total ergot alkaloids, R²_C for calibration was less than 0.55 and 0.96, R²_Cv for cross validation was less than 0.14 and 0.96, R²_p for external prediction was NA, for barley and wheat, respectively. In general, the PLS models developed showed relatively weak cross-validation performance. More efforts are required to explore the direct detection limit of the NIR and ATR- FT/MIR techniques for the quantification in different sample matrix.