Does massage help neck pain?

Neck pain is second only to back pain as a reason for booking, and the 2024 Cochrane review of it contains the single most useful finding in the massage evidence base. It is not the headline result. It is a subgroup analysis most summaries skip.

The short version

• Against placebo, massage made little to no difference to pain, function or quality of life at up to 12 weeks. Low certainty throughout.¹
• The review's own conclusion: "The contribution of massage to the management of neck pain remains uncertain given the predominance of low-certainty evidence in this field."¹
• The dose subgroup analysis found a clinically important difference favouring eight or more sessions over four weeks, of 30 minutes or longer each.¹
• Most of the placebo trials tested a single session, and the review downgraded its own evidence partly because of it.¹
• That makes the dose finding the best available argument for a course of treatment, with the caveat that a subgroup analysis is weaker evidence than a headline result.

What the 2024 review found

The Cochrane review updated in February 2024 included 33 randomised trials with 1,994 participants analysed. Participants were aged 18 to 70, 70% female, with mean pain of 51.8 out of 100 on a visual analogue scale. Most had non-specific neck pain (85%), with smaller groups having radiculopathy (6%) or cervicogenic headache (9%).¹

Against placebo at up to 12 weeks:

Two things about that table matter more than the numbers in it. The pain confidence interval crosses zero, meaning the result is compatible with massage doing nothing. And the function, treatment success and quality of life rows rest on one or two small studies each, which is why every row is graded low certainty.

The review was blunt about the trials themselves: selection bias in 82% and detection bias in 94%, with several using a placebo that may not have been credible and never testing whether blinding actually worked.¹

This is consistent with the wider field. Across 129 systematic reviews of massage for pain published from 2018 to 2023, no conclusion was rated high certainty.²

The dose finding, and why it is the interesting part

Now the part that is genuinely useful.

The review ran a subgroup analysis by dose and found a clinically important difference favouring a high dose, defined as eight or more sessions over four weeks, each 30 minutes or longer.¹

Set that against how the placebo trials were actually built. When the review explained why it downgraded the pain evidence, one reason given was indirectness, because "most trials in the placebo comparison used suboptimal massage doses (only single sessions)".¹

So the unimpressive headline is largely a measurement of what one massage does, and the review flags that as a limitation of the trials rather than a finding about massage. When the analysis separates out the studies that delivered a proper course, the picture changes.

Three honest caveats belong with it.

1. A subgroup analysis is weaker than a primary result. It is generated from the data after the fact and carries a higher chance of a false positive. It should be read as the most promising signal available, not as proof.
2. The overall certainty is still low. A dose finding inside low-certainty evidence does not become high-certainty evidence.
3. It has not been tested directly. The review's closing recommendation is that "focused planning for larger, adequately dosed, well-designed trials is needed".¹ Nobody has yet run the trial this finding asks for.

What it does establish is that the common test people apply, booking one massage and judging massage on it, is the same test the trials applied, and the review names it as a weakness of those trials.

What this evidence does not cover

Cervicogenic headache. Headache originating in the neck has its own literature and it is regularly borrowed to sell neck massage. It should not be. A 2022 review of manual and exercise therapy for cervicogenic headache found moderate to large short-term effects, but the sensitivity analysis restricted to low risk of bias trials shrank that to small effects, and both trials in that sensitivity analysis studied spinal manipulation, not massage.³ The stronger result belongs to a different intervention.

Neck pain with radiculopathy, meaning pain, numbness or weakness travelling down the arm from a nerve root, made up 6% of participants in the Cochrane review.¹ There is no separate evidence base for massage treating it. New or worsening arm weakness or numbness is a reason to see a doctor rather than book a massage.

Whiplash. Only one trial in the review involved whiplash.¹ There is not enough there to say anything either way.

Safety

The review found very low-certainty evidence on total adverse events, with the risk ratio essentially at no difference but a confidence interval so wide it establishes nothing (RR 0.99, 95% CI 0.08 to 11.55, from 2 studies).¹ The adverse events recorded were treatment soreness, sweating and low blood pressure. No data on serious adverse events was available, because the trials did not report it adequately.

The broader safety literature puts serious adverse events from massage at the level of "probably true rarities", most often associated with techniques other than standard Swedish massage and frequently delivered by people without training.⁴

Neck work deserves one specific note. Serious events in the literature on neck treatment are associated with high-velocity spinal manipulation, which is a chiropractic and osteopathic technique, not a massage one. Remedial massage does not involve neck manipulation.

Reading it honestly

Massage for neck pain sits in a common position: widely used, genuinely liked, and thinly evidenced. The reviewers say the contribution remains uncertain, and that is the accurate summary.

The most defensible version of the case for it is the dose one. If you are going to test whether massage helps your neck, the evidence suggests testing it properly, at around eight sessions of at least 30 minutes over four weeks, rather than booking one and drawing a conclusion. That is what the only positive signal in the review points at, and it is worth saying that this also happens to be the more expensive answer.

Which is exactly why it should come with its caveats attached, and they are above.

Frequently asked questions

Does massage help neck pain?

The evidence is uncertain. The 2024 Cochrane review of 33 trials found little to no difference against placebo for pain, function or quality of life at up to 12 weeks, all on low-certainty evidence. The one positive signal was a subgroup analysis favouring a higher dose of treatment.

How many massages do I need for neck pain?

The Cochrane subgroup analysis found a clinically important difference favouring eight or more sessions over four weeks, each 30 minutes or longer. That is the best available guide, though it comes from a subgroup analysis within low-certainty evidence rather than from a trial designed to test it.

Why did the research find massage did not work for neck pain?

Largely because most of the placebo-controlled trials tested a single session. The review named that as a reason it downgraded its own evidence, describing the doses used as suboptimal, and its dose analysis found a clinically important difference when a proper course was given.

Does massage help headaches from the neck?

The evidence usually cited for cervicogenic headache does not support massage specifically. When a 2022 review restricted its analysis to low risk of bias trials, both remaining trials studied spinal manipulation rather than massage.

Is massage safe for the neck?

For most people, yes. Recorded adverse events in the trials were soreness, sweating and low blood pressure. The serious events reported in the literature on neck treatment involve high-velocity spinal manipulation, which is not a remedial massage technique. See a doctor first for new or worsening numbness or weakness in an arm.

Should I get a massage or see a physiotherapist for neck pain?

There is no head-to-head evidence that settles this. If your neck pain comes with arm symptoms, follows an injury, or is not improving, start with a doctor or physiotherapist for assessment. If it is persistent non-specific neck stiffness and soreness, either is reasonable, and the dose evidence applies whichever you choose.