To optimize and establish the pretreatment method for aflatoxins (AFS) in grains, the method was optimized by using naturally contaminated grain samples. The grains were extracted by crushing, cleaned up by immunoaffinity column, and quantified by ultra performance liquid chromatography with fluorescence detector. The results showed that the extraction efficiency of 84% acetonitrile water was better than that of 70% methanol water, without soaking, and more than 99% of aflatoxin B1 could be extracted from the samples by vortex shaking for 20 min or high-speed homogenization for 2 min. The method has good accuracy and reproducibility. The results of the determination of aflatoxin B1 components in corn flour, brown rice flour, rice flour and other grain matrices of the national certified standards were within the range of standard values and their extended uncertainties, with a coefficient of variation of 5.2%, and the determination results were more satisfactory. Using this method to participate in the FAPAS (PT04359) proficiency testing, |Z|<2, and the results were satisfactory. Aflatoxin in grain by 84% acetonitrile water vortex extraction 20 min or homogenization extraction 2 min extraction efficiency and accuracy, the literature recommended the use of 70% methanol water extraction solution, and the use of ultrasound and other extraction methods may lead to low determination results, which cannot truly reflect the content of aflatoxin in the sample. Therefore, the unification of sample pretreatment methods is important for the accuracy and comparability of aflatoxin assay results in raw grains, and this study can provide reference for the selection of sample pretreatment methods for aflatoxin in grains.