Fusarium head blight (FHB) of wheat is caused by Fusarium graminearum and several other Fusarium species. FHB causes significant grain yield and quality losses due to accumulation of mycotoxins such as deoxynivalenol (DON) in infected kernels. Host resistance is considered the most practical and effective means of controlling FHB. DON levels in harvested grains are often used to assess FHB reaction of wheat varieties. A non destructive single kernel method was developed to estimate DON content in wheat grain samples using single kernel DON levels estimated by our single kernel near infrared (SKNIR) system. We estimated DON in 500-kernel grain samples from FHB screening trials conducted in 2010 and 2011 and compared those with DON determined by the gas chromatography-mass spectrometry (GC-MS) method. Results showed that DON levels of small kernel samples can be estimated by the SKNIR method using an estimate of single kernel weight and predicted DON levels. This method estimated DON levels in 90% of 160 samples within 6.7 ppm of DON levels determined by GC-MS. Single kernel DON analysis showed that DON content among DON containing kernels (DCK) varied considerably. Distribution of single kernel DON levels among kernels was used to compare Fusarium head blight (FHB) reaction of wheat varieties. Study of the DON distribution among all kernels as well as among DCK of a grain sample is helpful for an in-depth evaluation of FHB reaction of wheat varieties. The SKNIR single kernel DON analysis and study of single kernel DON distribution patterns demonstrated in this study may help wheat breeders to comprehensively screen varieties in relation to their resistance to spread of disease and resistance to DON accumulation.