Objective：To compare the image quality of gradual and non-gradual breath training protocols in gadoxetate disodium （Gd-EOB-DTPA） upper abdomen-enhanced magnetic resonance at hepatic arterial phase，combing with dilution method of 1∶1，so as to explore the most reasonable combination of breath training and injection program. Methods：A total of 95 patients underwent Gd-EOB-DTPA upper abdomen-enhanced MR imaging in our department from May to September 2019 with no significant difference in the population statistics were selected and divided into three groups：the group 1（30 patients were trained by gradual protocol with the breath time from 5 s，10 s，15 s to 20 s，as no ups and downs of whole abdomen for the standard），the group 2（33 pa－tients were trained by non-gradual protocol with the breath time of 20 s，as no ups and downs of whole abdomen for the stan－dard），and the group 3（32 patients were trained with the same protocol as the group 2）. The Gd-EOB-DTPA injection rate in the group 1 and the group 2 was 1 mL/s，with 20 mL normal saline flushing. The 1∶1 diluted with saline Gd-EOB-DTPA injection rate in the group 3 was 2 mL/s，with 20 mL normal saline flushing. All arterial phases were detected by MR fluoroscopic triggering technique. The scoring standard for image quality included：1= no artefact，2=mild and diagnosable artefact，3= moderate and diagnosable arte－fact，4=severe and diagnosable artefact and 5=undiagnosable. Imaging score of hepatic artery，portal vein，abdominal aorta and liver parenchyma was evaluated and the total score were calculated，and differences of the 3 group were analyzed by the Kruskal-Wallis test and Mann-Whitney test. Results：The mean score of hepatic artery，portal vein，abdominal aorta and liver parenchyma and the total score in the group 3 were slightly higher than those in the group 1 and the group 2（P=0.065，0.229，0.819，0.340，0.080）. The breath－ing artefact in the group 3 was significant smaller than that in the group 1 and the group 2（P=0.004，0.046）. The non-gradual diluted method had a smaller respiratory artefact. Conclusion：Holding breath for 20s in the group 3 is adopted and Gd-EOB-DTPA diluted 50% injection method can significantly decrease the respiratory motion artefacts.