We grouped hemiplegic patients with shoulder subluxation according to the presence of hemiplegic shoulder pain (HSP) and aimed to compare the difference in clinical and magnetic resonance (MR) arthrography findings between 2 groups in the effort to find factors possibly related to HSP. We performed a retrospective, cross-sectional analysis of all eligible inpatients treated at one university rehabilitation hospital. Clinical findings including passive range of motion (ROM), motor power, and spasticity and MR arthrography findings in the affected shoulder were compared between the 2 groups. Of 59 patients who met the criteria, 21 (35.6%) and 38 (64.4%) were classified as HSP group and no-HSP group, respectively. Limitation of ROM in flexion and rotator cuff atrophy significantly related with HSP (R² = 0.449, p < 0.05). Limitation of ROM in flexion and rotator cuff atrophy in MR arthrography was 1.1 and 6.4 times more likely associated with HSP, respectively. Additionally, spasticity in flexion correlated with limitation of ROM in flexion (p = 0.049), external rotation (p = 0.034), and with total limitation of ROM (p = 0.033). A significant correlation was evident between HSP and ROM of the shoulder. However, HSP did not correlate with MR arthrography findings, except rotator cuff atrophy.