<p>Cadmium (Cd) is a heavy metal that is highly toxic to the human body. Wheat grown in high-Cd soil can accumulate Cd in its kernels at a level exceeding international standards. Currently established methods to measure Cd content in grain are expensive and time-consuming and require skilled sample handling, highlighting the need for a quick and relatively simple method of determining the Cd concentration in wheat grain. The approach reported here predicted the Cd concentration in wheat grain samples by using partial least squares (PLS) regression analysis to associate near-infrared (NIR) spectrometry of grain samples with their corresponding Cd content as determined by chemical analysis. Four spectrum pre-processing methods and three wavelength ranges were compared; detrending of the spectra from 981 to 1095 nm was found to yield the best PLS regression performance in the cross-validation, with a root-mean-square error of 0.082 mg kg⁻¹ and a correlation coefficient of 0.952. Robustness of this method was validated by repeatedly partitioning one-fourth of the samples as the test dataset and then predicting their Cd content with the model developed from the remainder of the samples. NIR spectrometry can measure many wheat grain samples in a short time, making this method feasible for examination of harvested grains at farms and in screening for low-Cd-accumulating parental materials for breeding purposes.</p>