BackgroundPost-stroke shoulder pain (PSSP) is characterized by shoulder pain on the hemiplegic side, which can limit physical activity in patients with stroke. Acupuncture combined with rehabilitation training (AR) has been widely used in PSSP, but the evidence of its effectiveness is still unclear.ObjectiveThe study aimed to evaluate the effect and safety of AR vs. rehabilitation training (RT) alone on PSSP.MethodsWe searched PubMed, the Cochrane Library, the Chinese Biological Medicine Database (CBM), the Chinese Scientific Journal Database (VIP), China National Knowledge Infrastructure (CNKI), and the WAN FANG database for relevant studies from their inception to February 2022. Only randomized controlled trials (RCTs) comparing the effect of AR with RT alone on PSSP were considered. The primary outcome was shoulder pain. Secondary outcomes included upper limb motor function, activities of daily living (ADL), shoulder range of motion (ROM), and adverse events (AEs). Subgroup analysis and sensitivity analysis were also conducted. Quality assessment was implemented based on Cochrane risk of bias (ROB) criteria, which consist of seven items. When more than four items in a study were judged as low ROB, the overall quality of this study was considered low risk.ResultsA total of 40 studies were included in the qualitative analysis, and 35 (87.5%) studies with 2,554 patients were included in the meta-analysis. Of the 40 studies, 14 (35.0%) were of moderate-to-high quality. The meta-analysis results showed that AR is better than RT alone in reducing shoulder pain (MD −1.32, 95% CI −1.58 to −1.07), improving upper limb motor function (MD 6.81, 95% CI 4.95–8.67), ADL (MD 11.17, 95% CI 9.44–12.91), and shoulder ROM (internal rotation: MD 10.48, 95% CI 8.14–12.83; backward extension: MD 7.82, 95% CI 6.00–9.64; anteflexion: MD 12.88, 95% CI 5.47–20.29; external rotation: MD 11.40, 95% CI 6.17–16.64; abduction: MD 16.96, 95% CI 8.61–25.31) without obvious AEs.ConclusionAR may be better than RT alone for the improvement of shoulder pain, upper limb motor function, ADL, and shoulder ROM, without obvious AEs in patients with PSSP. However, considering the clinical and statistical heterogeneity, our findings need to be interpreted with caution. More rigorous RCTs in this area should be conducted in the future.Systematic review registration[www.crd.york.ac.uk], identifier [CRD42022326763].