ObjectivesTo overview the systematic reviews/meta-analyses (SRs/MAs) of effectiveness and safety of spinal manipulation for low back pain or neck pain. MethodsWe electronically searched databases including PubMed, EMbase, The Cochrane Library (Issue 1, 2015), CBM, CNKI, WanFang Data and VIP to collect SRs/MAs of spinal manipulation for low back pain or neck pain from inception to January 30th, 2015. Two reviewers independently screened literature, extracted data, and then AMSTAR tool was used to assess the methodological quality of included SRs/MAs. ResultsA total of 21 SRs/MAs were included. Twenty of them assessed the methodological quality of included original randomized controlled trials (RCTs) with different tools:2 used Jadad scale, 5 used PEDro scale, 6 used Cochrane bias risk assessment tool and 7 used other tools. The assessment results of AMSTAR tool suggested that:among 11 items, the item 1 of "Was an ‘a priori’ design provided" (18 SRs/MAs did not provide) and item 4 of "Was a list of studies (included and excluded) provided" (18 SRs/MAs did not provide) appeared to be the most problematic, followed by item 10 of "Was the likelihood of publication bias assessed" (14 SRs/MAs did not assess the publication bias) and item 11 of "Was the conflict of interest stated" (14 SRs/MAs did not provide the conflict of interest and 4 were incomplete). ConclusionThe methodological quality of included SRs/MAs is poor. The limited evidence showed that spinal manipulation is more effective for acute low back pain than chronic low back pain, and the short term effect is better than the long term one. Different spinal manipulation techniques have various effects but are all safe. Chiropractic manipulation may have the best effect. Due to the limitation of quality and quantity of included SRs/MAs, there may be potential bias in the above conclusion that needs more high quality studies to verify.