Exercise therapy for chronic neck pain: what the evidence shows

Why this question is worth asking carefully

If your neck has ached for months — stiff in the morning, tight by the end of a screen-heavy day, sore in a way that comes and goes but never quite leaves — you have a lot of company, and you have probably already been told to "do some exercises." It is one of the first things a physiotherapist, a doctor, or the internet will suggest. That makes it worth asking plainly: does exercise actually help long-standing neck pain, and if it does, how much, for how long, and does it matter which exercises? The honest answer is more interesting than a simple yes or no. Some kinds of exercise have reasonable evidence behind them for chronic neck pain; others, including the gentle stretching many people reach for first, have surprisingly little. And the overall quality of the research is a real part of the story, not a footnote. This piece lays out what the best current syntheses found — which exercise types, how large the effect, how confident researchers are, and how durable it looks — so you can have a better-informed conversation with a regulated health professional. We are not here to sell you on exercise or to talk you out of it; we are here to show the evidence as it stands.

What "chronic non-specific neck pain" actually means

Neck pain is usually defined simply: pain in the neck, with or without pain referred into one or both arms, lasting at least a day [1]. It becomes chronic when it persists — commonly counted from around three months onward. The word that does the most work here is non-specific. It means that after a proper assessment, the pain has not been traced to a specific structural cause such as a fracture, a tumour, an infection, or a compressed nerve root producing arm symptoms. Most ordinary, persistent neck pain falls into this non-specific category — which is precisely why treatment tends to focus on managing symptoms and restoring function rather than fixing an identified lesion.

It is also one of the most burdensome conditions there is. The Global Burden of Disease Study 2021 ranks neck pain among the leading musculoskeletal causes of disability worldwide, with the number of years people live with disability from it rising from about 11.5 million in 1990 to roughly 20.2 million in 2020, and no evidence that neck pain itself causes death [1]. Women are affected more than men, the burden peaks in mid-to-late working age, and cases are projected to keep climbing toward mid-century [1]. That scale — a common, disabling, rarely dangerous problem with no single cure — is exactly the setting in which a low-risk, widely available option like exercise draws so much attention.

What the evidence shows

The most established synthesis is a Cochrane systematic review by Gross and colleagues, which pooled the randomized trials of exercise for mechanical neck disorders [2]. Cochrane reviews sit at the top tier of evidence because of how rigorously they appraise each trial and grade overall confidence. Two things stand out from it — one encouraging, one cautionary.

The encouraging part is that specific, targeted strengthening did better than exercise in general. For chronic neck pain, strengthening exercises for the cervico-scapulothoracic region and the upper extremity — the muscles of the neck, shoulder blades, and upper back — produced a moderate-to-large reduction in pain immediately after treatment (standardized mean difference −0.71, 95% CI −1.33 to −0.10) [2]. Combined strengthening and stretching programmes showed a small-to-large beneficial effect on pain with a medium-sized improvement in function, and cervico-scapulothoracic strengthening and stabilization work also improved pain and function into the intermediate term [2][4]. Endurance training for the same regions produced slighter benefits [2].

The cautionary part comes in two forms. First, stretching on its own did not deliver. The review found that breathing exercises, general fitness training, and stretching alone showed little to no change in pain or function — in the reviewers' words, when only stretching exercises were used, no beneficial effect could be expected [2]. That is worth pausing on, because gentle stretching is often the first and sometimes the only thing people try. The evidence points toward loaded, targeted strengthening of the neck-and-shoulder complex as the component doing the work, not stretching by itself. Second, and framing everything above, Cochrane concluded that no high-quality evidence was found — certainty across the comparisons ranged from low to moderate [2][3]. So the fair reading is: specific strengthening looks helpful and is reasonable to use as part of routine care for chronic neck pain, but the confidence behind that conclusion is limited, not firm.

A more recent umbrella review — a systematic review and meta-analysis of systematic reviews by Rasmussen-Barr and colleagues — sharpens the picture for particular exercise types [4]. Compared with people who did no exercise, and measured in the short term, it reported:

• Motor control exercises (targeted training of the deep neck stabilizers and coordination): a large point estimate for pain (SMD −1.69, 95% CI −2.73 to −0.64) and for disability (SMD −2.26, 95% CI −3.38 to −1.39) [4].
• Yoga: pain SMD −1.32 (95% CI −1.84 to −0.80) and disability SMD −1.00 (95% CI −1.47 to −0.54) [4].
• Resistance training: a moderate benefit for pain (SMD −0.75, 95% CI −1.41 to −0.09), with no significant effect on disability [4].
• Traditional Chinese exercise (Tai Chi and Qigong): a smaller but consistent benefit for pain (SMD −0.63, 95% CI −0.95 to −0.32) and disability (SMD −0.39, 95% CI −0.65 to −0.13) [4].

Those point estimates look impressive, and some are genuinely large — but they should be read with three restraints the review itself carries. The confidence intervals are wide (motor control's, for instance, spans from a small to a very large effect), the certainty of evidence ranged only from low to high depending on the exercise and outcome, and the benefits were short-term: the authors found no high-quality trials establishing long-term effects [4]. A big number attached to a wide interval and a short follow-up is a promising signal, not a settled fact. Read alongside Cochrane's "no high-quality evidence" verdict, the honest summary is that several specific, active exercise approaches show real short-term benefit for chronic neck pain, with the strongest and best-studied signal around targeted strengthening and motor-control work — while the durability and the precise size of that benefit remain genuinely uncertain.

How it's studied — and why "exercise" is a slippery word

Three features of this research should temper how firmly anyone reads the numbers.

• "Exercise" is not one thing. The trials bundle together strengthening, motor-control training, stretching, endurance work, yoga, Tai Chi, general aerobic fitness, and postural drills — at different intensities, doses, and durations, supervised and unsupervised. When results this varied get pooled, an overall average can hide the fact that the components differ a lot: the same review that shows targeted strengthening helping also shows stretching alone doing little [2][4]. This is why the type of exercise matters as much as the instruction to exercise at all.
• The trials are often small and at risk of bias. A systematic review focused specifically on exercise dosage for chronic non-specific neck pain judged 23 of its 26 included trials to be at high risk of bias, with small samples and inconsistent reporting, and concluded that the optimal type and dose of exercise remains unknown and needs better trials to pin down [5]. Its most usable signal was directional — for motor-control exercise, the benefit appeared to strengthen as training frequency rose — rather than a precise prescription [5].
• You cannot blind an exercise trial. In a drug trial you can give an identical dummy pill; you cannot make someone unaware they were assigned to a strengthening programme. Most of these trials compared exercise against no exercise, which leaves open how much of the measured improvement is the exercise itself versus the attention, encouragement, and expectation that come with any supervised programme [2][4]. That is a real and unresolved source of uncertainty in this literature.

Taken together, these are the reasons the confidence ceiling stays at "low-to-moderate" even when individual point estimates look large. The signal is consistent enough to act on; the precision is not yet there.

What the guidelines say

Given a promising but imperfect evidence base, you might expect hedged guidance. Instead, clinical practice guidelines recommend exercise for persistent neck pain consistently — while being specific about what kind. The widely used Neck Pain clinical practice guidelines from the Orthopaedic Section of the American Physical Therapy Association (published in the Journal of Orthopaedic & Sports Physical Therapy) advise a multimodal approach for people with continuing neck pain and persistent impairments: manual mobilization techniques combined with exercise, and specifically strengthening, endurance, flexibility, coordination, postural, and functional exercises of the neck, shoulder-blade, and upper-body regions [6]. That aligns closely with what the trial evidence points to — targeted, active, loaded exercise of the neck-and-shoulder complex, delivered as part of a broader plan rather than a single stretch.

Why land on a firm recommendation when the effect sizes carry wide error bars? Because guideline panels weigh more than one number. They factor in that chronic non-specific neck pain has no cure and no low-risk fix, that active exercise is comparatively safe, inexpensive, and self-directed, that it supports general health, and that it is a reasonable first-line step before or alongside medication or more invasive options [1][6]. A recommendation to use exercise is a judgment about the overall balance of benefits, harms, and alternatives — not a claim that any one programme will reliably produce a large change in your neck. Reading it as the latter would overstate what the trials measured.

What this means for you

If you are living with chronic non-specific neck pain, a fair summary is this: the best current evidence suggests that specific, active exercise — particularly targeted strengthening and motor-control training for the neck, shoulder blades, and upper back — probably reduces pain and improves function in the short term, that the effect can be meaningful but is measured with limited certainty and wide margins, that stretching on its own has little support, that how long the benefit lasts is not well established, and that clinical guidelines still recommend supervised, multimodal exercise as a sensible first-line option [2][4][5][6]. None of that is a promise of relief, and none of it is a substitute for a proper assessment of your neck.

The right next step is a conversation with a licensed, regulated health professional — your physician, a physiotherapist, or another qualified clinician — who can examine your specific situation, rule out the less common problems that need different care, and help you decide which kind of exercise, at what dose and progression, makes sense for you and how it fits alongside your other options. Choosing and adjusting an exercise programme is exactly the individualized judgment that population averages cannot make for you — and it is worth getting that judgment from someone who has actually assessed you.