Start Here Explore Stack Verdict Skip List Supplements Subscribe
The VerdictMODERATE CONVICTION

Your neck isn't tight. It's getting the wrong instructions, and that's why stretching never worked.

Before anything else: if your head is drooping and you tire quickly, if your speech or swallowing has changed, or if this began within days of a new medication, do not do the test below. See a doctor today. Otherwise, check this: sit tall and rest a fingertip lightly on your cheek, your chin, or the back of your head. Not pushing. Just touching. Notice whether the pulling eases. If it does, that is called a sensory trick, and it is one of the two findings clinicians use to identify this condition. Book an appointment and use that exact word. Takes less than a minute. No equipment needed.

  1. What this actually is: a signal problem in the brain that shows up as a neck problem, which is why perfectly healthy muscles end up pulling all day.
  2. What most people get wrong: treating the tight-feeling muscle, when that muscle is following orders rather than misbehaving.
  3. Start here: check whether a light fingertip on your face eases the pulling, and if it does, get referred, because the treatment with the strongest evidence is one your physical therapist cannot give you.

Think of a thermostat wired to the wrong room. The heater is fine and the wiring is fine, but the signal telling it what to do is reading the temperature somewhere else entirely. Your neck muscles are the heater. They are healthy, they are strong, and they are doing exactly what they are told. The fault is in the signal, which is why every treatment aimed at the muscle has left the instructions untouched.

SH
Dr. Seth Holbrook, DPT — Doctor of Physical Therapy • Coach to 300+ clients
I built The Verdict to cut through recycled health advice and show what the evidence actually supports.

The Verdict · Physio

Cervical Dystonia

Also called spasmodic torticollis

A stiff neck that pulls itself to one side, over and over, in the same pattern. The muscles are healthy. The instructions reaching them are not.

Cervical Spine Conviction: Moderate Evidence reviewed 2026

What Works

Ranked by the strength of the evidence behind each one, not by how often it gets offered.

Dark cinematic study of cervical musculature under dramatic lighting

Tier 1 — Strong evidence

Botulinum toxin type A injection HIGH

Given by a neurologist, injected into the muscles doing the pulling to quieten the signal reaching them. Across 9 randomised trials and 1,144 patients it beat a dummy injection by 8.09 points on the standard severity scale (95% CI 6.22 to 9.96), which the reviewers themselves convert to 18.4% improvement from where people started.

The honest caveat, carried with the number. 7 of those 9 trials had already excluded people who responded poorly to the treatment. Only 1 of the 9 was independently funded. Every one measured a single injection at week 4. And the improvement threshold researchers established for this same scale is 11.9 points, with a second study putting it at 8. The pooled result sits below one and at the edge of the other, so this is a treatment that clearly works better than nothing and should still be measured on you individually.

Tier 2 — Moderate evidence

Physical therapy alongside the injections MODERATE

Aimed at pain, which is where its evidence actually is. Pooled reduction of 5.00 points on the pain scale (95% CI 6.26 to 3.74) when added on top of the injections.

That pooled figure came from 2 studies out of 14 the reviewers looked at, and they state plainly that the treatments varied too much for them to recommend any specific programme. So the direction is trustworthy and the detail is not.

Exercise Prescription

There is no proven dose for this condition, and pretending otherwise would be the easiest lie on this page. The content below is what an international panel of physical therapists agreed on. The schedule is clinical judgment. Your own therapist should change it to fit you.
ExerciseWhat you doSets × RepsFrequencyHow it should feel
Guided head repositioningSitting tall, gently guide your head back toward the middle using your own hand as a light cue, then let go. You are practising the position, not forcing it2 × 8 slowDailyEffort and awareness, never a fight. Stop if it triggers a stronger pull
Light sensory cueingRest a fingertip lightly on your cheek, chin, or the back of your head. Notice the pulling ease. Hold for a few slow breaths3 × 20-30 secDaily, and any time symptoms spikeNo pain at all. A light touch, not a push
Postural settingLengthen through the top of your head and let your shoulders drop. Hold, breathe, release5 × 10 sec2-3× dailyNo pain
Active neck movementSlow controlled turns and tilts through the range you have. Not stretching into the end range1 × 6 each wayDailyMild effort only
Relaxation practiceSlow breathing or progressive muscle relaxation, 10 minutes1 sessionDailyNot applicable

Two things worth knowing before you start. Expect mild muscle soreness: in the trial that measured it, two-thirds of people doing corrective exercises reported it, and there were no adverse events in either group. And that last row is not filler. In the one trial that compared corrective exercise against relaxation alone, the two produced the same result.

Tier 3 and the other options

Deep brain stimulation MODERATE
The guideline position is that it is a good option, particularly for this condition, after medication or botulinum toxin have failed. Less effective where the dystonia is secondary to something else. It needs a specialist multidisciplinary team, and it is a referral-level decision well outside physical therapy.

Anticholinergic medication (trihexyphenidyl) LOW
One trial, 66 people, funded by the maker of the competing treatment. Botulinum toxin came out 2.5 points ahead on severity, with 31 side effects against 76 on the tablets. A Cochrane review resting on a single trial is a statement about how thin the literature is, not about the drug.

Sensory retraining as treatment LOW
Listed as essential by the physical therapy consensus panel. No trial establishes a dose, a protocol, or an effect size for it. In this literature the sensory trick earns its keep mainly as a way of identifying the condition.

What doesn't work

  • Treating the pulling muscle as a tight muscle. Stretching, needling and deep tissue work aimed at the hypertrophied side address a consequence. No trial here tested or supported a muscle-length approach.
  • Physical therapy instead of a referral. It has never been tested on its own in this condition. Every measurable benefit in the literature was produced on top of the injections.
  • Choosing one physical therapy programme over another. 96 patients, 12 months, specialised against regular: no difference on the main outcome (P=.326). The specialised programme did win on how people rated their own health and on cost, which is a real argument, just not the one usually made for it.
  • Presenting active exercise as better than relaxation. In the one trial that compared them fairly, they matched.

Red Flags

Read this before anything else. Most neck problems are not urgent. A small number of the things that look like this one are. They are sorted below by how fast you need to act.

Dark cinematic anatomical study of the cervical region

Same-day medical care

  • The pulling started within days of beginning a new medication, especially anti-sickness medication (metoclopramide) or an antipsychotic (risperidone, aripiprazole). This is called an acute dystonic reaction. In a reported series every case resolved within 30 minutes of the right injection, and the authors note it can reach life-threatening severity if left.
  • Your head is drooping forward, your neck tires quickly, or your speech or swallowing has changed. Myasthenia gravis and motor neurone disease both present this way, and both were deliberately included as comparison cases in the study that validated how this condition is identified.

Contact your doctor this week

  • Swallowing has become difficult after a botulinum toxin injection cycle. This is a recognised side effect and it goes back to the clinician who injected you, not to your physical therapist.

Get assessed rather than treated

  • It started before you were 30, or a relative developed a movement problem young. There are specific tests and a trial of a particular medication that should happen before anyone starts a rehabilitation programme. A reported case was misdiagnosed and operated on before a low dose of that medication improved it.
  • Months of competent neck treatment have changed nothing. Not an emergency. It is the single most common reason this condition goes unrecognised for years.

Referral routes: neurology, specifically a movement disorder service, for the diagnosis and for treatment access. Same-day GP or emergency care for a drug reaction. Urgent neurology for a drooping head with fatigue. Eye clinic if the tilt disappears when one eye is covered.

This page is for information. It cannot rule anything out, and it is not a substitute for being examined.

Return to Training

This is a condition managed rather than cured, so these are criteria for going back to full unrestricted loading, not for being fixed. No trial defines return-to-activity criteria for this condition, so every item below is clinical judgment and is labelled as such.

The training lever here is sustained fixed head position under load, not total volume. Nothing in this evidence supports resting or deloading, and no trial reported an adverse event from active exercise. Heavy overhead work, bar placement in front squats, and long isometric holds are the practical candidates for modification. Worth knowing: stress is the most frequently reported aggravator in the dystonia pain literature, so a training block that coincides with a rough patch at work may be reading as a training problem when it is not.

Conviction

MODERATE and deliberately stratified, because the confidence is very different depending on which claim you are asking about.

What would change my mind on the injection's effect size

A trial of botulinum toxin that does not exclude people who previously responded poorly, reporting results across a full injection cycle rather than at week 4 of a single session, and analysed against the 11.9-point meaningful-change threshold rather than against zero. That would tell us whether the 8.09-point pooled figure is a real ceiling or an artefact of picking favourable patients and measuring early.

What would change my mind on physical therapy

A multicentre trial of at least 150 adults with confirmed cervical dystonia, comparing a properly written physical therapy programme against a credible control with the same amount of contact time, split by whether the person is receiving injections, with the standard severity scale as the main outcome at 6 and 12 months. A positive result in the group not receiving injections would be the first evidence this condition has a rehabilitation answer that does not depend on the needle. That trial does not exist.

Go Deeper

Most stiff necks are exactly what they look like. The ones that aren't cost people years. The Verdict breaks down one condition a week, with the evidence graded honestly, including where it's thin.

Join The Verdict — free
The Full Picture — Anatomy, Diagnosis & Evidence

What's Actually Going On

Dark cinematic anatomical rendering of deep cervical musculature

Cervical dystonia produces repetitive, patterned movements or postures that pull the head away from neutral. The muscles doing the pulling are structurally normal. They are receiving abnormal instructions from a disordered sensorimotor network, and the visible neck posture is the output of that disorder rather than its cause.

That one distinction governs everything else on this page. Lengthening the muscle that is pulling does not change what is pulling it.

The deviation is usually a mix of rotation, sideways tilt, and forward or backward pull, and most people present with a combination rather than a pure pattern. Because the pattern is individual, expert consensus specifically recommends assessing someone in several different positions rather than one, and separating the steady pull from the jerky, tremulous component.

There is more going on than the posture. A systematic review of 20 studies found subtle difficulties with processing speed, memory, visual and spatial skills, planning, and reading social situations, while language and attention were relatively spared. That matters for how a home programme is designed, not for how the condition is identified.

Pain is close to universal in adult-onset dystonia. In the study validating a dystonia-specific pain scale, 90.0% of a Canadian group and 96.3% of an Italian group reported it, with stress named as the most frequent aggravator and rest and stretching as the most common relievers. Worth stating precisely: those two groups had facial and arm dystonia, not neck. The scale itself was built and validated in cervical dystonia, which is why the study exists, but those particular percentages are not neck percentages.

How to Identify It

Dark cinematic clinical assessment scene of the head and neck

One study did the work that matters here. It filmed 43 people with the condition and 41 comparison cases, and the comparison group was built deliberately from the things this gets confused with: isolated head tremor, chorea, tics, drooping heads caused by myasthenia gravis or motor neurone disease, orthopaedic and rheumatological neck problems, and head tilts driven by an eye problem. Three independent raters scored a 6-item guideline against a senior neurologist's diagnosis.

  • Item 1 — is the movement repetitive and patterned? A head that deviates from neutral the same way every time, rather than randomly. Core feature
  • Item 2 — is there a sensory trick? Light contact to the cheek, chin or back of the head temporarily reduces the pull. Core feature
  • Items 3 to 6 — the four features that should be absent. Each one points toward a condition that mimics this rather than toward the condition itself.

Best combination: Sensitivity 96.1% Specificity 81%, meaning it catches roughly 96 in 100 who have it and correctly clears roughly 81 in 100 who do not.

The interesting part is the item they had to throw out. "Can you hold your head straight if you try?" is the intuitive question, and including it made the accuracy worse. People with the mimicking conditions can often partly suppress the movement too, and plenty of people with this condition can as well. Asking it does no harm. Weighting it does.

The Debate

The guidelines have expired, and consensus moved into the empty space

European guideline, 2011 · North American guideline, 2016

Botulinum toxin is first-line. Deep brain stimulation comes after medication or injections have failed. Both are the formal authority for this condition.

vs

Two expert consensus panels, both 2026

Both recommend integrating physical therapy and structured non-drug care. Neither is a clinical guideline, and both exist because nothing else does.

This is not a contradiction, it is a vacuum. The European guideline is 15 years old. The North American one carries the word RETIRED in its own title on the medical index, formally withdrawn with nothing replacing it. When guidelines expire without successors, consensus panels fill the space and then get quoted in the language of guidance. Worth knowing which you are reading: the only document in existence that tells a physical therapist what to assess and treat in this condition reached agreement among 10 people in its first round and 9 in its second, and its own authors describe it as expert opinion to be tested in future trials.

The first-line treatment against its own scale's meaningful-change threshold

Cochrane review, 2020 · 9 trials · 1,144 patients

Botulinum toxin beat placebo by 8.09 points (95% CI 6.22 to 9.96), measured at week 4 after a single session.

vs

Two threshold studies · 304 and 479 patients

The change that patients themselves register as meaningful on that same scale is 11.9 points in one study and 8 in the other.

Same scale, both thresholds worked out in people receiving the same treatment, so the comparison is fair and nobody in the literature seems to make it. The pooled effect lands below one threshold and at the lower edge of the other. This does not mean the treatment fails. It means "proven to work" and "a difference you would notice" are two different sentences, and the second one has to be checked on the individual in front of you. It is only fair to add what travels with that number: a single injection, measured at four weeks, in trials where 7 of 9 had already excluded people who responded poorly.

Honest Limitations

Limitation 1 — the trials studied a different patient

The research finding: physical therapy added to injections reduced pain by a pooled 5.00 points.

The real-world gap: the largest trial required people to be stable on injections for a full year before they were randomised. The pooled effect is defined as therapy used as an addition to injections. The person who walks into a physical therapy clinic with an intractable stiff neck is undiagnosed, unreferred, and receiving no injections at all. That person is outside every trial in this literature.

What that changes: the first visit is a recognition problem, not a rehabilitation problem. The add-on evidence only becomes yours to use once the referral has happened.

Limitation 2 — the effect shrinks as the study design tightens

The research finding: the biggest reported benefit is a pain change of 13.35 points with added therapy, against a comparison group whose pain got worse by 6.95 over the same period.

The real-world gap: that trial was unblinded with a no-therapy comparison, so all the extra attention and expectation loads onto the treated side, and a comparison group deteriorating like that is the fingerprint of it. Blind the assessment and give the comparison group relaxation, and the difference is 1.9 points with a range running from 9.0 in favour to 5.2 against. Compare one programme against another in 96 people over a year and the main outcome comes back at P=.326. Three trials, three tightenings, three shrinkages.

What that changes: expect a real but modest contribution, mostly on pain. Do not promise anyone that the posture will be corrected.

Limitation 3 — there is no dose, and the reviewers say so themselves

The research finding: the reviewers state that the variety in the type and duration of physical therapy did not allow them to recommend any specific type. The pooled estimate came from 2 studies out of 14.

The real-world gap: a clinician looking for sets, reps, frequency or progression criteria will find none, and the natural move is to import them from ordinary mechanical neck pain, where they are well established and where the underlying problem is a completely different thing.

What that changes: use the agreed content, and say out loud to the patient that the schedule is judgment rather than evidence. That is what the exercise table on this page does.

The Nuance

Dark cinematic anatomical study contrasting cervical structures

The honest position on cure. This condition does not have a cure pathway in the evidence, and no document in the literature claims one. It has a first-line injection with genuine trial support whose pooled size sits below the threshold people notice, an add-on therapy with a modest pain benefit pooled from two studies, and a surgical escalation for those the injection fails. The realistic conversation is about management and expectation.

Surgery against conservative care, stated honestly. No retrieved study reports what proportion of people recover fully with conservative management, and no usable natural-history or remission rate exists for a general population with this condition. On the surgical side, a cost analysis found deep brain stimulation clinically better than a sham procedure in one trial per condition, without a retrievable proportion of responders. The cost figures are real and the success percentages simply are not there. Anyone quoting you one has got it from somewhere other than this literature.

One name, two entirely different conditions. The old term for this is spasmodic torticollis, and "torticollis" on its own overwhelmingly refers to congenital muscular torticollis, a tightness of one neck muscle in babies. They share nothing but a discarded word. It is worth knowing because it is the reason searching for information about this condition turns up so much material about infants, and it is a genuine source of confusion for people trying to read about their own diagnosis.

Sources

Dealing with something specific?

Every pain and rehab verdict, evidence-scored: what actually speeds recovery, what to skip, and when to get it checked.

Browse Pain & Rehab verdicts
Or find your lane in 2 questions

Get weekly evidence-based rehab verdicts

Physio conditions reviewed against clinical evidence. What works, what doesn't, and what to do — from a practising physiotherapist.

Subscribe free

Want a coach, not just research?

The Verdict is built by the same team behind Precision Metrics — a physique and health coaching practice with 300+ clients coached. Dr. Seth Holbrook, DPT and Luke Holbrook lead the coaching.

Book a free consultation

Related free research

Pain & Rehab
Accessory Deep Peroneal Nerve — The Verdict
 Pain & Rehab
Dorsal Foot and Ankle Bone Spurs as a Nerve-Compression Cause — The Verdict
 Pain & Rehab
Charcot-Marie-Tooth Disease — The Verdict

There are 500+ more inside

Conviction-scored verdicts on supplements, nutrition, training, physio, and recovery.

Explore all Get weekly verdicts