Background Motor imagery-based brain–computer Interface (MI-BCI) utilises electroencephalographic signals from imagined limb movements for control commands, enabling real-time interaction with external devices. Clinical trials suggest its potential for upper limb recovery post-stroke. This study aims to systematically evaluate the effectiveness of MI-BCI on upper limb motor function in patients with post-stroke hemiplegia. Methods A comprehensive search was conducted across PubMed, Cochrane Library, Web of Science and Embase through March 2026 for randomised controlled trials assessing MI-BCI effects on upper limb motor impairments post-stroke. A systematic review and meta-analysis were performed. Results The analysis included 27 studies for systematic review, with meta-analyses encompassing 24 studies involving 846 participants. MI-BCI training showed statistically significant improvements in measures of isolated limb movement and fine motor control, specifically the Fugl-Meyer Assessment for Upper Extremity (s.m.d. 0.31, 95% CI: 0.18–0.45; I² = 17%) and the Wolf Motor Function Test (s.m.d. 0.45, 95% CI: 0.24–0.66; I² = 39%). Conclusions MI-BCI training can improve upper limb motor function, particularly for isolated movements and fine motor control, in stroke patients. Its effects on complex functional activities, activities of daily living and spasticity were not significant in the current evidence. Due to heterogeneity in control conditions (including sham interventions) across studies, the current evidence does not support definitive conclusions regarding its superiority over standardised traditional rehabilitation.