There is a rapidly growing body of evidence for the application of virtual reality (VR) in pain management, however, with varying effectiveness. Little is known about patient-related and VR-related factors affecting efficacy of VR. A systematic review and meta-analysis was performed including 122 randomised controlled trials (9138 patients), reporting on subjectively reported pain scores comparing an immersive VR intervention to a non-VR control group. Virtual reality significantly reduced pain in the pooled analysis (standardized mean difference = -0.65, 95% CI -0.76 to -0.54, P < 0.001). Subgroup analyses showed no significant differences between type of pain, ie, VR effects were similar in acute, chronic, and procedural pain conditions. Univariate and multivariate meta-regression analyses were performed to investigate the effect of intervention, patient, and pain characteristics on VR. Virtual reality effectively reduced pain, especially in patients reporting moderate to severe pain and in younger subjects. Studies comparing VR with a control group receiving no distraction methods were associated with higher effect sizes. The effect of VR was not related to a specific frequency or duration of use. Type of software and interaction level were related to VR effects in the univariable, but not in the multivariable, meta-regression analysis. Heterogeneity was considerable for all meta-analyses, and risk of bias was moderate to high in most included studies. Studies on mechanisms behind VR analgesia in younger patients and patients reporting moderate to severe pain are recommended to confirm our hypotheses while taking into account risk of bias and the comparator. Optimal application of VR using treatment modules for long-term pain conditions are an important issue for future research.