Background: Numerous studies have reported that pelvic functional training is beneficial for improving low back pain (LBP) in patients with lumbar disc herniation (LDH) after lumbar surgery and that fat infiltration of the lumbar multifidus muscles (LMMs) is one of the most important reasons for residual LBP after surgery. However, little is known about the exact relationship among residual LBP, spinopelvic parameters and the area of fat infiltration in LMM after lumbar surgery. This study aimed to confirm the relationship among residual LBP, spinopelvic parameters and the area of fat infiltration in LMM and to investigate why pelvic functional training can relieve pain symptoms in patients with LDH after lumbar surgery.Methods: One hundred forty-three patients with LDH were involved in this study. Clinical data were collected from a system of digital medical records, including age, gender, course, and weight. On the MRI images, the cross-sectional areas (CSAs) of bilateral fat infiltration in the LMM were measured using a picture archiving and communication system (PACS). On the X-ray, sacral slope (SS), pelvic tilt (PT) and pelvic incidence (PI) were also measured by PACS. Pearson correlation analysis was applied to analyse the differences between CSA of fat infiltration in LMM and spinopelvic parameters, and ROC curves were used to reflect the degree of fat infiltration in LMM with spinopelvic parameters.Results: One hundred and twenty-five patients met the inclusion criteria. SS and PI were positively correlated with CSA of fat infiltration in LMM at L3-4 and L4-5 (p < 0.01). At L4-5, SS and PI demonstrated significant positive correlation with the CSA of fat infiltration in the LMM (0.5 < | r | < 0.8). PI also exhibited a significant positive correlation with VAS (0.5 < | r | < 0.8), but SS had a low correlation with VAS (0.3 < | r | < 0.5). At L4-5, only the PI had a significant ROC curve (AUC = 0.836) with a cut-off point score of 0.556 and sensitivity and specificity values of 62.8% and 87.4%, respectively. However, at L3-4 and L5-S1, the AUCs of the ROC curves were all < 0.7 for SS, PT and PI. Conclusions: The wider pelvic anterior tilt, the more severe fat the infiltration in LMM. Residual LBP can be relieved by spinopelvic correction training potentially due to the improvement of fat infiltration in LMM.