Background: Chronic pain is one of the most feared symptoms in people with cancer. Although music therapy is frequently used for pain management in cancer care, there is a lack of knowledge of underlying mechanisms. Yet, it is well accepted that knowledge of mediators is essential for optimization of psychosocial treatment interventions. We conducted a multi-site randomized controlled trial to examine mediators hypothesized to account for the pain-reducing effects of interactive music therapy (IMT) in adults with advanced cancer. Methods: This study included adults with advanced cancer (stage III/IV) who had moderate to severe chronic pain caused by 1) locally advanced cancer that had extended to other organs/soft tissue or was impinging on/eroding the bone, 2) bone metastasis, or 3) soft tissue metastasis. We excluded adults with primary central nervous system (CNS) tumor, CNS metastatic disease, or hematologic malignancies. We randomized 92 participants to 6 individual IMT (n = 45) or social attention control (SAC) (n = 47) sessions. The mediators (anxiety, mood, pain-related self-efficacy and emotional support) and pain outcomes (pain intensity and pain interference) were measured at baseline, week 4, and week 6 using self-report measures. We included treatment expectancy, music reward, and adult playfulness as moderators. Follow-up interviews were conducted to examine congruence between the hypothesized mediation model and participants’ explanations of how IMT impacted their pain experience. Our target sample size was 100 to achieve 80% power to detect a medium effect at α = 0.05 and 28% attrition. We used ITT analyses and the mediation model per Preacher and Hayesto estimate the total and specific indirect effects of each hypothesized mediator individually. Interview transcripts were analyzed using thematic analysis. Results: Participants were 56±12 years old. Most were female (72%), white (44%) or Black (39%), and had stage IV cancer (75%). The most common types of cancer were breast (46%) and lung (13%). Self-efficacy was found to be a significant mediator for the effect of IMT on pain intensity (indirect effect ab = 0.79, 95% CI 0.01 to 1.82) and pain interference (indirect effect ab = 1.16, 95% CI 0.02 to 2.51), while anxiety, mood, and perceived support were not. The quantitative results were supported by the qualitative findings. None of the proposed moderators were found to be significant. However, we found that the mediating effect of pain-related self-efficacy on pain intensity and pain interference was significantly moderated by baseline pain interference. Conclusions: This study suggests that the impact of music therapy on chronic pain management is mediated through self-efficacy. This knowledge can help further optimize music therapy interventions for chronic pain management in people with advanced cancer by capitalizing on teaching music-based self-management techniques. Clinical trial information: NCT03432247.