Background: A wide variety of pharmacological and non-pharmacological interventions are available for non-specific low back pain (NSLBP). However, there are few head-to-head trials available of non-pharmacological interventions for NSLBP, and it is not clear which intervention should be the first choice. Objective: We aimed to conduct an overview of Cochrane systematic reviews (CSRs) on NSLBP by summarizing both direct and indirect evidence to identify the most effective non-pharmacological treatment for low back pain. Methods: We searched the Cochrane Library for reviews examining interventions for NSLBP. We extracted data on type of intervention, patients characteristics and outcome measures, and calculated standardized mean differences of treatment effect. Active interventions were compared using direct comparison (DC) and adjusted indirect comparison (AIC). Results: We included 3 reviews involving 109 trials examining non-pharmacological interventions (massage, exercise therapy and acupuncture) for NSLBP. The common outcomes in these reviews were pain, functional status and work absenteeism, and we chose to evaluate pain as our primary outcome. The direct comparison trials showed that massage was statistically significantly better than exercise therapy or acupuncture. The indirect comparisons also indicate massage may be better than acupuncture or exercise, although the difference between massage and acupuncture was statistically non-significant by the indirect comparison (Figure & Table 1). The discrepancy between DC and AIC estimates was statistically significant when massage was compared with acupuncture (p = 0.01) and not significant for the comparison of massage and exercise. Conclusions: We illustrate the feasibility of conducting an overview of reviews in a complex area, and succinctly summarize the relative effects of competing interests in an accessible format. Massage appears to be the first-line option for patients with NSLBP as compared to exercise or acupuncture.