Fusarium head blight (FHB) is a major cereal disease caused by a complex of Fusarium species. These species vary in importance depending on climatic conditions, agronomic factors or host genotype. In addition, Fusarium species can release toxic secondary metabolites. These mycotoxins constitute a significant food safety concern as they have health implications in both humans and animals. The Fusarium species involved in FHB differ in their pathogenicity, ability to produce mycotoxins, and fungicide sensitivity. Accurate and exhaustive identification of Fusarium species in planta is therefore of great importance. In this study, using a new set of primers targeting the EF1α gene, the diversity of Fusarium species on cereals was evaluated using Illumina high-throughput sequencing. The PCR amplification parameters and bioinformatic pipeline were optimized with mock and artificially infected grain communities and further tested on 65 field samples. Fusarium species were retrieved from mock communities and good reproducibility between different runs or PCR cycle numbers was be observed. The method enabled the detection of as few as one single Fusarium-infected grain in 10,000. Up to 17 different Fusarium species were detected in field samples of barley, durum and soft wheat harvested in France. This new set of primers enables the assessment of Fusarium diversity by high-throughput sequencing on cereal samples. It provides a more exhaustive picture of the Fusarium community than the currently used techniques based on isolation or species-specific PCR detection. This new experimental approach may be used to show changes in the composition of the Fusarium complex or to detect the emergence of new Fusarium species as far as the EF1α sequence of these species show a sufficient amount of polymorphism in the portion of sequence analyzed. Information on the distribution and prevalence of the different Fusarium species in a given geographical area, and in response to various environmental factors, is of great interest for managing the disease and predicting mycotoxin contamination risks.