A fully cited, honestly caveated reference on chiropractic and spinal manipulation safety: mild side-effect rates, serious adverse event data from 154 trials, the stroke question, and what the numbers can and cannot prove.
Search "chiropractic success rate" and you will find confident numbers: 89 percent this, 90 percent that, almost always on a clinic's own website and almost never with a citation you can check. Safety gets the same treatment from the opposite direction, with scare pieces built on single case reports. This page is our attempt to do the harder and more useful thing: lay out what the actual research says about how safe chiropractic adjustments are, with every figure sourced, and with the limitations stated plainly rather than buried. We are a chiropractic practice. We also think the honest version of this data is more reassuring than the marketing version, precisely because you can verify it.
This is the most studied part of chiropractic safety, and the numbers are consistent across decades and countries. In a large prospective study, Senstad and colleagues followed more than a thousand chiropractic patients across roughly 4,700 treatments and found that about 55 percent reported at least one unpleasant reaction during their course of care. The reactions were overwhelmingly mild: local discomfort in the treated area was the most common, followed by headache and tiredness. Critically, most reactions appeared the same day, and roughly three quarters had resolved within 24 hours.
Cagnie and colleagues found nearly identical numbers a few years later, with about 60 percent of patients reporting a reaction, most beginning within four hours of treatment and most gone within a day. The UCLA Neck Pain Study, which looked specifically at neck care, reported the same pattern: increased pain or headache was the typical reaction, and it was transient.
| Study | Population | Patients with any reaction | Typical timeline |
|---|---|---|---|
| Senstad 1997 (Spine) | ~1,058 patients, 4,712 treatments | ~55% | Onset same day, ~74% gone within 24 h |
| Cagnie 2004 (Manual Therapy) | 465 patients | ~60% | Most onset within 4 h, most gone within 24 h |
| Hurwitz 2005 (Spine, UCLA neck pain) | Neck-pain patients in an RCT | ~30% of visits with any reaction | Transient increased pain or headache |
None of this is unique to chiropractic. Studies that gave patients a sham (pretend) treatment recorded plenty of soreness reports too, which tells you how much of the "reaction" is simply a body being handled and moved. We wrote a patient-facing companion to this section in our article on why you feel sore after an adjustment.
The headline number that anchors this page comes from a 2023 systematic review published in BMJ Open by Gorrell and colleagues. Looking across 154 randomized controlled trials that together enrolled 7,518 participants who received spinal manipulation, the reviewers found that no serious adverse events were reported in any of them. Earlier systematic reviews reached compatible conclusions. Gouveia and colleagues (2009) reviewed the safety literature and found serious complications to be rare, and a 2007 review by Ernst concluded that serious adverse events were reported only sporadically, mostly as isolated case reports rather than in controlled studies.
Taken at face value, that is a strong safety signal: thousands of people manipulated under controlled conditions, zero serious harms. But taking it purely at face value would be exactly the kind of overclaiming this page exists to avoid, so here is the asterisk.
The same BMJ Open review that produced the reassuring headline also documented a reporting problem serious enough that the authors called for a standardized system to fix it:
The honest reading: serious adverse events from spinal manipulation are clearly uncommon, the controlled evidence contains none, and the controlled evidence is also not large enough or consistent enough to put a precise number on a rare risk. Both halves of that sentence are true, and pages that only tell you the first half are selling, not informing.
The most serious event associated with neck manipulation, and the one that generates the most fear, is a stroke caused by a tear in one of the arteries running through the neck, called a cervical artery dissection. It is worth walking through carefully, because the evidence here is better than most people assume and it does not say what either the boosters or the alarmists claim.
The landmark study is a population-based analysis by Cassidy and colleagues, published in Spine in 2008, covering roughly 100 million person-years of data in Ontario. It found that people who had this specific type of stroke were more likely to have seen a chiropractor beforehand, which sounds damning until you see the control comparison: they were also more likely to have seen a primary care physician beforehand, by a similar margin. Since a family doctor's visit does not mechanically cause an artery to tear, the most plausible explanation is that people in the earliest stage of a dissection develop neck pain and headache, the classic warning symptoms, and seek care for those symptoms from whichever provider they use, shortly before the stroke completes. The care is a marker of the developing problem, not its cause.
Later studies using different populations reached the same conclusion. Whedon and colleagues (2015) examined Medicare beneficiaries aged 66 to 99 and found no excess stroke risk after chiropractic manipulation compared with a primary care visit. Kosloff and colleagues (2015) analyzed commercial and Medicare Advantage populations and likewise found no significant association suggesting excess risk. And a 2016 systematic review and meta-analysis by Church and colleagues concluded there was no convincing evidence for a causal link between chiropractic neck manipulation and cervical artery dissection, while noting the underlying data quality is low and cannot completely exclude a small risk.
Cervical artery dissection is rare to begin with, the best available research does not show that neck manipulation causes it, and the association that does appear in the data is best explained by patients seeking care for the warning symptoms of a stroke already in progress. That is genuinely reassuring. What we will not tell you is that the risk is provably zero, because the studies themselves are honest that their data quality is low and a very small risk cannot be fully ruled out. This is also why a careful practitioner screens for the warning signs of dissection before manipulating a neck, and refers rather than adjusts when they appear.
For low-back manipulation, the serious complication of concern is cauda equina syndrome, a compression of the bundle of nerve roots at the base of the spinal canal. In reviews of manipulation safety it appears only as rare case reports, and its usual cause is a large disc herniation rather than manipulation itself. It is a genuine medical emergency whenever it occurs, which is why its warning signs, new loss of bladder or bowel control, numbness in the groin or saddle area, and progressive leg weakness, belong on every patient's radar regardless of what preceded them. We describe those warning signs in plain language in our articles on back pain after an adjustment and herniated discs. The practical safety point is the same as for the neck: a proper examination screens for the red flags before any manual treatment, and those findings send a patient to the emergency department, not the adjusting table.
If you read only one section of this page, read this one, because it is the finding that most distinguishes honest safety data from marketing. Across the safety literature, the recurring theme is not that spinal manipulation has been proven dangerous or proven perfectly safe. It is that the studies have measured harm inconsistently. The 2023 BMJ Open review found most trials either did not report adverse events or did not define them. Reviews going back years have made the same complaint. This is why you should be skeptical of any precise safety percentage, in either direction: the denominator and the definitions are often missing.
The constructive response, which the BMJ Open authors called for explicitly, is a standardized system for defining, classifying, and reporting adverse events in manipulation trials, so that future numbers actually mean something. Until that exists, the most accurate summary of chiropractic adjustment safety is a plain-language one rather than a single statistic: mild transient reactions are common and expected, serious events are rare, the rare-event risk cannot be pinned to a precise number with current data, and honest practice manages that uncertainty with screening, appropriate technique, and prompt referral when red flags appear.
Statistics describe populations. What protects an individual patient is the process in the treatment room, and it is worth knowing what that looks like so you can expect it:
A screening examination before treatment. History, neurological and orthopedic testing, and imaging when indicated exist largely to catch the small number of presentations that should not be manipulated at all, from suspected fracture to the warning signs of dissection or cauda equina. At our offices this screen is the first visit's main job, and when it turns up something outside our lane, the same-day move is a referral, not an adjustment.
Technique matched to the patient. A high-velocity adjustment is one tool among many. Gentler instrument-assisted and drop-table techniques, mobilization, and soft-tissue work accomplish similar goals for patients who are older, acutely irritated, or simply prefer a lighter approach. We describe these options in our overview of chiropractic techniques for back pain.
Informed consent and feedback. Knowing that mild soreness is likely, and telling your provider how you responded, is not a formality. It is how the plan gets tuned and how an unusual reaction gets caught early.
Safety is never absolute, only relative to the alternatives for the same problem. For the mechanical back and neck pain that brings most people to a chiropractor, the common alternatives are medication and time. Anti-inflammatory drugs carry well-documented gastrointestinal, kidney, and cardiovascular risks with regular use, and opioids carry risks of dependence and overdose that are the subject of a national public-health crisis. Against that backdrop, the 2017 JAMA systematic review of spinal manipulation for acute low-back pain found modest benefit with the main harm being transient muscle soreness, and the U.S. National Center for Complementary and Integrative Health describes serious complications as rare.
This is why the American College of Physicians, in its 2017 clinical practice guideline, recommends non-drug treatments including spinal manipulation as a first-line option for low-back pain, ahead of medication. The point is not that manipulation is risk-free. It is that its risk profile, mild transient soreness for most and rare serious events, compares favorably with the pharmacologic options for the same conditions, which is a genuinely different message from either "perfectly safe" or "dangerous."
Every quantitative claim on this page is sourced to a systematic review, population study, or randomized trial listed in the references below, each verifiable through PubMed, the Cochrane Library, or the publishing journal. We prioritized systematic reviews and large population studies over individual reports, stated the evidence-quality limitations alongside the figures rather than in a separate disclaimer, and preserved the original studies' own hedges, including where authors rated their evidence as low quality or noted that their data could not exclude a small risk. This page reflects the literature as of its last-reviewed date at the top; the safety evidence evolves, and we revise the page when significant new reviews appear. Nothing here is medical advice for an individual, and it does not replace an in-person evaluation.
Mild, temporary side effects such as local soreness, stiffness, headache, or tiredness are common, affecting roughly half of patients at some point, and usually resolve within a day or two. Serious adverse events are rare: across 154 randomized trials with 7,518 people receiving spinal manipulation, none were reported. That figure is reassuring but limited, because only 61 percent of those trials tracked adverse events and the sample is too small to measure a truly rare risk.
The best population research does not show that neck manipulation causes stroke. A landmark 2008 study found that stroke patients were no more likely to have seen a chiropractor beforehand than to have seen a primary care doctor, which suggests people seek care for the early symptoms of a stroke already in progress rather than the care causing it. Later studies agree, and a 2016 meta-analysis found no convincing evidence of a causal link, while noting the data cannot fully exclude a small risk.
Local soreness or discomfort in the treated area is the most common, followed by headache and tiredness. Studies consistently find these reactions in roughly 55 to 60 percent of patients at some point during care, usually beginning within a few hours of treatment and resolving within 24 to 48 hours. They are most common after the first visit or two.
For mechanical back and neck pain, the main harm from spinal manipulation reported in trials is transient muscle soreness, while common drug alternatives carry their own documented risks: anti-inflammatories affect the gut, kidneys, and heart with regular use, and opioids carry dependence and overdose risk. The American College of Physicians recommends non-drug options including spinal manipulation as first-line care for low-back pain, ahead of medication.
Because the studies have measured harm inconsistently. A 2023 review found most manipulation trials either did not report adverse events or did not define what counted as one. This makes precise safety percentages unreliable in either direction and is why the most accurate summary is a plain-language one: mild reactions are common, serious events are rare, and the rare-event risk cannot be pinned to an exact number with current data.
This article is for general education and is not a substitute for an individual evaluation. External links are provided for reference and do not imply endorsement.
