<![CDATA[Background: Virtual Reality (VR) has emerged as a contemporary modality in neurorehabilitation, utilizing interactive environments and multisensory feedback to support neuroplasticity and motor learning. However, the influence of stroke phase and VR immersion level on therapeutic effectiveness has not been systematically synthesized. Objective: To systematically synthesize the effectiveness of VR-based interventions on upper limb motor function after stroke and critically evaluate the moderating effects of stroke phase and VR immersion level on rehabilitation outcomes. Methods: A systematic review and meta-analysis were conducted in accordance with PRISMA guidelines, with searches across major scientific databases. Experimental studies, including randomized controlled trials (RCTs) and quasi-experimental studies, assessing upper limb motor function using the Fugl-Meyer Assessment–Upper Extremity (FMA-UE) were included. Risk of bias was evaluated using the Cochrane Risk of Bias tool. A random-effects model was applied to calculate pooled Mean Differences (MDs), with subgroup analyses based on stroke phase and VR immersion level. Results: Eleven studies (10 RCTs and 1 quasi-experimental study) involving 388 participants were included. VR-based interventions demonstrated a statistically significant improvement in upper limb motor function compared with conventional therapy (MD = 8.97; 95% CI 6.11–11.82; p < 0.00001). Significant effects were consistently observed in both the chronic phase (MD = 9.96) and the subacute phase (MD = 8.36). Both non-immersive and immersive VR yielded clinically meaningful benefits, with no significant difference between immersion levels (p = 0.97). Conclusion: Virtual Reality based neurorehabilitation significantly enhances post-stroke upper limb motor recovery, with consistent effectiveness regardless of stroke phase or device immersion level. These findings support VR as a flexible and clinically applicable adjunct to standard rehabilitation, although standardized intervention protocols and long-term outcome evaluations remain necessary.]]>
