Objective: To explore the correlation between the ultrasonic manifestations of shoulder soft tissue lesions and visual analogue scale (VAS) in patients with hemiplegic shoulder pain (HSP), and to provide a new method for quantitative assessment of the shoulder lesion. Methods: A total of 62 patients with hemiplegia were enrolled, according to the VAS scale score, they were divided into shoulder pain group (37 cases) and non-shoulder pain group (25 cases). All of them were evaluated by Brunnstrom, modified Ashworth, VAS. All shoulders were examined by musculoskeletal ultrasound, and analyzed the correlation between ultrasound abnormalities and VAS in shoulder pain group. Results: In the shoulder pain group, the incidence of the long head of biceps tendon, supraspinatus lesion and subacromial-subdeltoid bursa effusion in the shoulder pain group was higher than that in the non-shoulder pain group [51.3% (19/37) vs 28% (7/25), 56.8% (21/37) vs 24% (6/25), 40.5% (15/37) vs 16.0%(4/25); P<0.05)], and the difference was statistically significant. The thickness of the inferior capsule and the distance from the acromion to the greater tubercle were greater than that of the non-shoulder pain [(3.69±1.03) mm vs (3.14±0.91) mm, (30.59±2.58) mm vs (26.98±2.84) mm; P<0.05)], and the difference was statistically significant. The correlation between abnormal sonographic findings and VAS correlation showed that the long head of biceps tendon lesion, the supraspinatus, and subdeltoid-subacromial bursa effusion, the thickness of the joint capsule, the distance from the acromion to the greater tubercle were all positively correlated with VAS in the shoulder pain group (r=0.54, r=0.60, r=0.38, r=0.55, r=0.61; P<0.05), and the difference was statistically significant. Conclusion: There is a correlation between ultrasound abnormalities and shoulder pain. Musculoskeletal ultrasound can be used as a new method for quantitative assessment of soft tissue lesions of the shoulder.