BACKGROUND: Back and neck pain are important health problems with serious societal and economic implications. Conventional treatments have been shown to have limited benefit in improving patient outcomes. Complementary and Alternative Medicine (CAM) therapies offer additional options in the management of low back and neck pain. Many trials evaluating CAM therapies have poor quality and inconsistent results. OBJECTIVES: To systematically review the efficacy, effectiveness, cost-effectiveness, and harms of acupuncture, spinal manipulation, mobilization, and massage techniques in management of back, neck, and/or thoracic pain. DATA SOURCES: MEDLINE, Cochrane Central, Cochrane Database of Systematic Reviews, CINAHL, and EMBASE were searched up to 2010; unpublished literature and reference lists of relevant articles were also searched. study selection: All records were screened by two independent reviewers. Primary reports of comparative efficacy, effectiveness, harms, and/or economic evaluations from randomized controlled trials (RCTs) of the CAM therapies in adults (age ≥ 18 years) with back, neck, or thoracic pain were eligible. Non-randomized controlled trials and observational studies (case-control, cohort, cross-sectional) comparing harms were also included. Reviews, case reports, editorials, commentaries or letters were excluded. DATA EXTRACTION: Two independent reviewers using a predefined form extracted data on study, participants, treatments, and outcome characteristics. RESULTS: 265 RCTs and 5 non-RCTs were included. Acupuncture for chronic nonspecific low back pain was associated with significantly lower pain intensity than placebo but only immediately post-treatment (VAS: -0.59, 95 percent CI: -0.93, -0.25). However, acupuncture was not different from placebo in post-treatment disability, pain medication intake, or global improvement in chronic nonspecific low back pain. Acupuncture did not differ from sham-acupuncture in reducing chronic non-specific neck pain immediately after treatment (VAS: 0.24, 95 percent CI: -1.20, 0.73). Acupuncture was superior to no treatment in improving pain intensity (VAS: -1.19, 95 percent CI: 95 percent CI: -2.17, -0.21), disability (PDI), functioning (HFAQ), well-being (SF-36), and range of mobility (extension, flexion), immediately after the treatment. In general, trials that applied sham-acupuncture tended to produce negative results (i.e., statistically non-significant) compared to trials that applied other types of placebo (e.g., TENS, medication, laser). Results regarding comparisons with other active treatments (pain medication, mobilization, laser therapy) were less consistent Acupuncture was more cost-effective compared to usual care or no treatment for patients with chronic back pain. For both low back and neck pain, manipulation was significantly better than placebo or no treatment in reducing pain immediately or short-term after the end of treatment. Manipulation was also better than acupuncture in improving pain and function in chronic nonspecific low back pain. Results from studies comparing manipulation to massage, medication, or physiotherapy were inconsistent, either in favor of manipulation or indicating no significant difference between the two treatments. Findings of studies regarding costs of manipulation relative to other therapies were inconsistent. Mobilization was superior to no treatment but not different from placebo in reducing low back pain or spinal flexibility after the treatment. Mobilization was better than physiotherapy in reducing low back pain (VAS: -0.50, 95 percent CI: -0.70, -0.30) and disability (Oswestry: -4.93, 95 percent CI: -5.91, -3.96). In subjects with acute or subacute neck pain, mobilization compared to placebo significantly reduced neck pain. Mobilization and placebo did not differ in subjects with chronic neck pain. Massage was superior to placebo or no treatment in reducing pain and disability only amongst subjects with acute/sub-acute low back pain. Massage was also significantly better than physical therapy in improving back pain (VAS: -2.11, 95 percent CI: -3.15, -1.07) or disability. For subjects with neck pain, massage was better than no treatment, placebo, or exercise in improving pain or disability, but not neck flexibility. Some evidence indicated higher costs for massage use compared to general practitioner care for low back pain. Reporting of harms in RCTs was poor and inconsistent. Subjects receiving CAM therapies reported soreness or bleeding on the site of application after acupuncture and worsening of pain after manipulation or massage. In two case-control studies cervical manipulation was shown to be significantly associated with vertebral artery dissection or vertebrobasilar vascular accident. CONCLUSIONS: Evidence was of poor to moderate grade and most of it pertained to chronic nonspecific pain, making it difficult to draw more definitive conclusions regarding benefits and harms of CAM therapies in subjects with acute/subacute, mixed, or unknown duration of pain. The benefit of CAM treatments was mostly evident immediately or shortly after the end of the treatment and then faded with time. Very few studies reported long-term outcomes. There was insufficient data to explore subgroup effects. The trial results were inconsistent due probably to methodological and clinical diversity, thereby limiting the extent of quantitative synthesis and complicating interpretation of trial results. Strong efforts are warranted to improve the conduct methodology and reporting quality of primary studies of CAM therapies. Future well powered head to head comparisons of CAM treatments and trials comparing CAM to widely used active treatments that report on all clinically relevant outcomes are needed to draw better conclusions.