Background <p>Intra-articular corticosteroid injection in combination with rehabilitation is a prevalent treatment modality for moderate-to-severe shoulder stiffness. However, this approach frequently results in delayed and incomplete recovery. Conversely, arthroscopic capsular release is associated with inherent surgical risks. Manipulation under anaesthesia (MUA) presents a minimally invasive alternative; however, concerns regarding intravenous anaesthesia and iatrogenic rotator cuff injury have served to limit its adoption. This randomised controlled trial aimed to evaluate the hypothesis that an Enhanced Recovery After Manipulation (ERAM) protocol incorporating an ultrasound-guided superior trunk brachial plexus block, manipulation under anaesthesia (MUA), an intra-articular corticosteroid injection and standardized rehabilitation is more effective than a corticosteroid injection and standardized rehabilitation.</p> Methods <p>In this prospective, single-center, parallel-group superiority trial, 112 patients with moderate-to-severe shoulder stiffness were randomized 1:1 to the ERAM protocol (<i>n</i> = 56) or intra-articular compound betamethasone injection plus standardized rehabilitation (<i>n</i> = 56). The primary outcome was the Constant-Murley shoulder function score. Secondary outcomes included visual analog scale (VAS) pain, passive range of motion (ROM) scores (forward flexion, abduction, external rotation, and internal rotation), and Patient Global Impression of Change (PGI-C), assessed at baseline, 1 week, 4 weeks, 8 weeks, and 6 months. Safety was evaluated with ultrasound and MRI at 8 weeks using the Snyder classification. Between-group comparisons used independent t-tests or Wilcoxon rank-sum tests; within-group changes were analyzed by repeated-measures ANOVA with Bonferroni correction.</p> Results <p>In total, 106 patients (53 per group) completed the 6-month follow-up (5.4% attrition). Baseline characteristics were comparable between groups. The ERAM group demonstrated significantly superior Constant-Murley scores, VAS pain scores, passive range of motion (composite score, forward flexion, abduction, external rotation, and internal rotation), and PGI-C ratings compared with the control group at every post-treatment time point (all <i>P</i> &lt; 0.05; exact p-values reported in Table&#xa0;2). Shoulder function in the ERAM group at 1 week was comparable to that of the control group at 6 months, with the greatest ROM gains occurring in forward flexion and abduction and no evidence of recurrent capsular contracture throughout follow-up. No severe adverse events occurred. One ERAM-group patient developed transient deltoid weakness attributed to perineural adhesions involving the axillary nerve, which resolved spontaneously within one day. Ultrasound and MRI at 8 weeks confirmed intact rotator cuff structures with unchanged Snyder grades in all patients.</p> Conclusion <p>For patients with moderate-to-severe shoulder stiffness, including frozen shoulder with or without partial-thickness rotator cuff tears (Snyder grades 1–2), the ERAM protocol substantially shortens the clinical course and achieves accelerated, durable functional recovery compared with corticosteroid injection and standardized rehabilitation. The protocol is compatible with an ambulatory care model and demonstrates a favorable safety profile.</p> Trial registration <p>This trial was registered at the (<a href="https://www.chictr.org.cn/">https://www.chictr.org.cn/</a>), (ChiCTR2400083123) on 16/4/2024.</p>