Objective To evaluate the diagnostic accuracy and repeatability of ACR TI-RADS classification for beginners.Methods Thyroid nodules confirmed by pathology after operation were collected from July 2018 to January 2020 in our hospital for thyroid ultrasound examination (with complete and qualified imaging data), ultrasoundguided fine needle aspiration biopsy (FNA) cytology examination. Through the imaging data that meet the requirements, the beginners were tested, trained and retested circularly. The results of each stage were recorded and compared. Results At each time point, the coincidence rate between the diagnosis results of senior doctors and the pathological results was 90.0%, 84.0% and 88.0%, respectively. The kappa consistency coefficient of diagnosis results of 3 novices and senior doctors was statistically significant and showed an increasing trend with time. The Kappa consistency coefficient of the diagnosticresults and pathological results of the three novices at the third time point was higher than that at the first time point. The correlation coefficients were 0.903, 0.926 and 0.872 in the same group at two different time points, and the correlation coefficients were statistically significant. Conclusion After shortterm training and learning, beginners can master the classification of ACR TI-RADS skillfully, with good diagnostic accuracy and repeatability.