Occipital Neuralgia

The VerdictLOW CONVICTION

A nerve runs through your neck muscles to your scalp.

Press firmly at the base of your skull, about two-thirds of the way from the middle out toward your ear, on the sore side. If that fires off the shooting pain rather than just feeling tender, that is your nerve. Then compare the skin on that patch of scalp against the other side. If the shooting pain reproduces and that patch of skin feels different, numb, or oddly sore, book an appointment this week. That combination is the finding most often missed.

What this actually is: sharp, electric, shooting pain from the base of the skull over one side of the scalp, usually with a patch of skin that has gone numb or oddly tender.
What most people get wrong: nearly half the people who have this also genuinely have migraine, so it is usually "and", not "instead of", and treating one while ignoring the other is the commonest reason people stay stuck.
Start here: get pressure off the back of your head. Pillow height, hats, headrest, hair tied tight on that side. It costs nothing and it is the one change nobody makes.

Think of a cable threaded through a bundle of ropes. In almost everyone, this nerve passes straight through the neck muscles rather than around them, so when those muscles tighten, the cable gets squeezed. But here is the part that changes everything: surgeons have cut this nerve, and the patients reliably lost feeling in their scalp while the pain often stayed. That tells you the alarm is not only coming from where you feel it. Part of it is coming from further up the line, which is why loosening the muscles is a reasonable place to start and not a guarantee.

Before anything else. If this pain came on suddenly and severely, especially after an injury or neck manipulation, or if it is the worst headache of your life, or you have weakness or trouble walking, go to A&E now. If you are over 50 with new scalp tenderness and jaw pain when chewing, see a doctor this week. Do not do the test below.

Press firmly at the base of your skull, two-thirds of the way from the middle out toward your ear, on the sore side. Then compare that patch of scalp against the other side.

If pressing fires off the shooting pain rather than just feeling tender, and that patch of skin feels numb or oddly sore, book an appointment this week. That combination is the finding most often missed.

Takes less than 2 minutes. No equipment needed.

The Verdict

A nerve runs through your neck muscles to your scalp. Irritate it and you get electric shocks, not aching.

Think of a cable threaded through a bundle of ropes. In almost everyone, this nerve passes straight through the neck muscles rather than around them, so when those muscles tighten, the cable gets squeezed. But here is the part that changes everything. Surgeons have cut this nerve, and the patients reliably lost feeling in their scalp while the pain often stayed. The alarm is not only coming from where you feel it, which is why loosening the muscles is a reasonable place to start and not a guarantee.

What this actually is: sharp, electric, shooting pain from the base of the skull over one side of the scalp, usually with a patch of skin that has gone numb or oddly tender.
What most people get wrong: nearly half the people who have this also genuinely have migraine, so it is usually "and", not "instead of", and treating one while ignoring the other is the commonest reason people stay stuck.
Start here: get pressure off the back of your head. Pillow height, hats, headrest, hair tied tight on that side. It costs nothing and it is the one change nobody makes.

Best for

Adults with sharp electric pain from the base of the skull over one side of the scalp, especially with a numb or over-sensitive patch, and a normal examination everywhere else.

Skip if

Your pain came on suddenly and severely, you have any weakness or trouble walking, you are over 50 with new scalp tenderness and jaw pain when chewing, or you have had surgery on your upper neck. Those need assessment, not exercises.

Want the full evidence? Keep scrolling

What Works

Read the grades before the treatments. This condition has no strongly-evidenced physical therapy treatment, and saying so is more useful than promoting something into a tier it did not earn.

Dark cinematic rendering of the posterior neck musculature and occipital region

Tier 1 — Strong Evidence

None exists for any hands-on or exercise treatment. The strongest evidence in this condition belongs to a procedure, and it is only moderate.

The best-evidenced option, and it is a referral

Greater occipital nerve block MODERATE

1 to 2 mL of local anaesthetic, ultrasound-guided at the second neck vertebra where available rather than by landmark at the base of the skull. Measurable within 30 minutes, with the difference from the landmark technique still there at 4 weeks. Side effects uncommon and mainly mild across 2,250 injections in six countries. It is a treatment, not a test.

Tier 2 — Moderate Evidence

Pulsed radiofrequency MODERATE

A specialist procedure. 3 cycles of 120 seconds per nerve. Beat steroid injection on average occipital pain at 6 weeks and held through 6 months in the only blinded head-to-head trial in this condition (Cohen 2015, 81 patients).

Conservative rehabilitation, guided by your symptom pattern MODERATE as reasoning

Neck and shoulder-girdle loading, hands-on work at the base of the skull, movement and posture retraining, education, and gradually reducing the scalp's sensitivity. The anatomy puts the nerve inside working muscle in 94 to 95 percent of people, so treatments that change tension and load tolerance there have a plausible target. First in the treatment sequence, second in the evidence ranking. Those are different rankings.

Tier 3 — Emerging and Thin

Neural mobilization LOW

Daily for 2 weeks in the only protocol ever tested. 68 percent rated "effective" in 22 unblinded patients, scored on a 1994 administrative criterion, and it did worse than the same protocol combined with electroacupuncture. This is the entire randomized evidence base for hands-on treatment in this condition, worldwide.

TENS LOW

A small device sending a mild electrical current through the skin. Settings were never reported. It is the only home treatment with any research specific to this condition, and that research is graded very low quality.

Occipital nerve stimulation, for cases that fail everything else LOW

All 15 studies behind the neurosurgical guideline are its weakest evidence class. Reported response is 63 to 100 percent, which is a range too wide to plan around. Ask for the complication rate in the same sentence: lead movement or fracture in 13 percent, local pain in 7.3 percent.

Exercise Prescription

Read this before you start. These are a cautious, low-risk starting point aimed at the right tissue. No research study has ever tested an exercise programme for this condition, so nobody, including your physical therapist, can tell you these specific movements will fix it. We use them because they are low-risk and target the right tissue, and we change them based on your response.

Chin nod

Lie on your back, knees bent. Gently nod your chin toward your chest as if saying a small "yes", without lifting your head off the floor.

3 × 10, holding 5 seconds · Daily · Gentle effort at the front of the neck. Stop if it triggers the shooting pain.

Suboccipital release

Lie on your back with a tennis ball or towel roll under the base of your skull on the sore side. Rest there and breathe.

60 to 90 seconds · Once or twice daily · A deep ache is fine. Sharp shooting pain means come off it immediately.

Upper trapezius and levator stretch

Sit tall. Take your ear toward the opposite shoulder, then add a small turn of the chin toward the same armpit. Light hand guidance only.

3 × 30 seconds each side · Daily · A stretch down the side and back of the neck. Never force it.

Scapular setting and band rows

With a light resistance band, draw the shoulder blades gently down and back and pull the band toward you. Keep the neck relaxed.

3 × 12 · 3 to 4 × per week · Effort in the mid-back. The neck should stay quiet.

Neck rotation, active range

Slowly turn your head to look over each shoulder, as far as is comfortable. Move smoothly, do not push into the pain.

2 × 10 each way · Daily · Stiff at worst, never sharp.

Review at 4 weeks. That is a clinical checkpoint, not a validated timeline. No change at 4 weeks means review the diagnosis, not push harder.

What Doesn't Work

Using the nerve block as proof of the diagnosis. It relieves migraine, cervicogenic headache, tension-type headache and cluster headache too.
Treating a co-existing migraine as the whole story. 46 percent have both.
Cutting the nerve to stop the pain. Numbness became near-certain and the pain rate did not change.
Acupuncture presented as established. Two reviews got near-identical results and the more recent team declined to endorse its own.
"Posture correction" with no dose, progression or review point. No parameters exist, so nobody can tell an inadequate dose from an ineffective treatment.
Upper neck manipulation. Never tested here, and the region's main red flag is a blood vessel.
Routine scans. Imaging rules other things out. It does not confirm this.

Red Flags

Occipital pain shares its presentation with conditions that are urgent. Screen before you treat.

Dark cinematic rendering of the upper cervical spine and skull base

Go to A&E or call emergency services now

Pain came on suddenly and severely, especially after an injury or neck manipulation
It is the worst headache of your life
Weakness, numbness or clumsiness in the arms or legs, or trouble walking
Slurred speech or double vision

See a doctor this week

Over 50 with new scalp tenderness, jaw pain when chewing, or any change in vision. This is a different condition that can threaten sight
Pain getting steadily worse rather than coming and going
Feeling generally unwell, feverish, or losing weight without trying
The numb patch is spreading, or the pain is moving into the face

Get assessed rather than treated

Previous surgery on the upper neck, or known instability there
A condition affecting the joints or connective tissue

A&E for sudden severe pain or acute neurological signs. Urgent doctor or eye specialist for suspected giant cell arteritis. Neurology for progressive or unusual presentations. Spinal surgery for suspected instability or pain after neck surgery.

Return to Training

Keep training. Nothing suggests loading harms this and nothing suggests rest fixes it. Change two things: keep the neck neutral, and get direct pressure off the back of the head.

Immediately: stop behind-the-neck pressing and pulldowns, heavy shrugs with a forward head position, end-range neck extension under load, and any headgear pressing on the back of the skull. Move overhead pressing to a neutral or slightly inclined position with a braced neck and less load. Keep the movement, cut the weight. Everything else continues.

Weeks 1 to 2: symptoms up to a mild background ache during and after training are acceptable. Reproducing the shooting pain is the stop signal, and it is a different signal from ordinary training discomfort. Keep your training frequency and cut load rather than the reverse.

Weeks 3 to 6: progress load on neutral-neck movements first, then reintroduce range. If the shooting episodes increase for more than 3 consecutive days, drop back to the last load that was quiet and hold there for a week.

Every criterion below is a clinical checkpoint. No return-to-activity criteria have ever been validated for this condition.

No shooting or electric occipital pain during any training movement for 2 consecutive weeks
Full, pain-free active neck rotation and extension
Overhead pressing at previous working weight with a neutral neck, with no symptoms during or in the 24 hours after
Scalp sensitivity settled enough that a pillow, hat or hair tie is tolerated
Background occipital ache absent or minimal on a normal training day
Any red flag symptom that appeared during the episode has been medically cleared

Conviction

LOW Endpoint-stratified. Some parts of this are solid. The parts that would tell you what to do are not.

The anatomy is HIGH. Two independent cadaver series agree on where the nerve runs and where it can be squeezed.

The clinical picture is MODERATE. Pooled from 15 clinic studies and 579 patients, with considerable variation between them.

How common it is: NO USABLE EVIDENCE. No population study exists anywhere. Clinic estimates range from 0.6 to 24.4 percent, a forty-fold spread.

Nerve block for short-term relief: MODERATE. Nerve block as a diagnostic confirmation: EVIDENCE AGAINST, not merely untested.

Exercise dose, progression, frequency, duration, return-to-activity criteria and any threshold for meaningful change: NO EVIDENCE.

What would change my mind about the nerve block not being diagnostic

A study enrolling consecutive occipital-pain presentations, applying the standard criteria, the nerve block and a bedside test battery, then following each patient to a clinically adjudicated final diagnosis at 12 months. That would establish for the first time whether the block has any ability to tell these conditions apart. Until an independent standard exists, further accuracy studies validated against the block will keep producing numbers that describe agreement rather than accuracy.

What would change my mind about hands-on and exercise treatment

A randomized trial of at least 120 adults meeting the standard criteria, allocated to a defined 8-week progressive neck and shoulder-girdle loading programme with neural mobilization versus a credible sham hands-on treatment, with headache days per month at 12 weeks as the primary outcome and a threshold for meaningful change registered in advance. Nothing smaller settles it, because the entire current evidence base is 22 unblinded patients.

Go Deeper

Most headache advice is written as if the diagnosis were settled. This one is not, and the honest version is more useful. The Verdict scores the evidence on one health question every week, free.

Get The Verdict free

The Full Picture — Anatomy, Diagnosis & Evidence

What's Actually Going On

Dark cinematic rendering of the greater occipital nerve pathway through the posterior neck

The greater occipital nerve comes off the second neck vertebra, hooks around a small deep muscle, then travels upward and pierces through muscle and fascia before it reaches the scalp. It passes through the neck extensor muscles, not around them.

In 100 nerves dissected from 50 cadavers, 95 percent pierced semispinalis capitis and 94 percent turned around the lower edge of obliquus capitis inferior (Huanmanop 2021). A separate dissection of 41 cadavers found four main shapes with 18 variations, and defined six separate points where the nerve could be compressed (Saglam 2024). So the entrapment story is anatomically sound, and the exact level differs between people and even between the two sides of the same person.

But the denervation data points somewhere else. Cutting the nerve during upper neck surgery produced numbness at odds ratio 178.6 and left the rate of occipital neuralgia unchanged at odds ratio 1.44 (Badhiwala 2017). Burning or cutting the nerve root produced numbness in 21 of 21 and 48 of 55 patients in the two arms of a 2026 comparison (Huang 2026). Denervation reliably removes feeling and unreliably removes pain, which is not what a purely mechanical squeeze would predict. The first look at human nerve tissue in this condition supports that: a specific population of pain-modulating cells was depleted in patients compared with controls (Wang Q 2026).

How to Identify It

Dark cinematic rendering of the skull base and suboccipital region

The diagnosis rests on the quality and time course of the pain, plus the sensory change, not on where it hurts.

Pooled from 15 clinic studies and 579 patients (Melchior 2025): unilateral in 81 percent, stabbing in 59 percent, severe in 54 percent, greater occipital nerve territory in 98 percent, and numbness or reduced sensation in 73 percent. Onset is typically in the fifth decade, and 73 percent are women. A history of neck trauma is present in 30 percent and a co-existing migraine in 46 percent, so neither rules it out.

Occipital Nerves Applied Strain test catches ~81 in 100 · clears only ~18 in 100

That is the test's performance against one reference standard; against the nerve block it catches about 94 in 100 and clears about 46 in 100 (Samolsky Dekel 2023, 163 consecutive patients). It can contribute to ruling the condition out. It cannot rule it in.

Nerve percussion at the occipital groove accuracy never published

It is still the highest-yield thing you can do at the bedside, alongside comparing scalp sensation side to side. Nobody has ever measured how well it performs.

Dark cinematic rendering contrasting cervical and cranial pain referral regions

Telling it apart: cervicogenic headache is steady, movement-provoked and comes with a restricted neck. Tension-type headache is bilateral, pressing and dull with no sensory change. Migraine lasts hours with nausea and light sensitivity, and often sits alongside this rather than instead of it. Cervical radiculopathy follows a dermatome and can affect strength. Any weakness at all takes the patient out of this diagnosis and into a medical assessment.

The Debate

Does a nerve block confirm the diagnosis?

Standard diagnostic criteria, in routine use

Temporary relief from a numbing injection around the affected nerve is treated as confirming occipital neuralgia. It is also the reference standard used to validate the one bedside test this condition has.

Gul 2017 (RCT, N=44) · Ulutas 2026 (N=2,219) · Fabry 2024

The same injection beat saline in chronic migraine at one, two and three months in a placebo-controlled trial. Across 2,219 patients in six countries it is the dominant procedure in migraine, tension-type headache and cluster headache, and used less in the neuralgias. A review across five headache conditions reports 60 to 90 percent efficacy in all of them.

A test that relieves everything it is meant to distinguish between confirms nothing. Use the block as a treatment. Diagnose on the electric quality, the burst-like time course and the sensory change. Every accuracy statistic in this condition was validated against this standard and inherits the problem.

Is acupuncture established here?

Zheng 2022, 15 studies

Pooled odds ratio 5.40 for treatment success and a large drop on the pain scale, both highly significant.

Yun 2020, 11 randomized trials

Near-identical numbers, odds ratio 4.96, and the authors concluded their own results were inconclusive.

Yun is right. The underlying trials share one origin, no blinding, an active drug comparator rather than a dummy, and a four-category administrative success score. Pooling trials that share a single systematic bias produces a tight range around a biased number, and the tightness reads as precision. Two reviews of the same flawed literature agreeing is not replication.

Honest Limitations

Nobody knows how common this is, within a factor of forty

A 2025 systematic review searched from 1988 to 2024 and found no population-based studies at all. Every figure comes from headache and facial-pain clinics assembled by different referral filters, so the 0.6 to 24.4 percent range is selection, not measurement error. Clinical intuition about how often this walks through the door has nothing behind it.

The reference standard is not specific, so the accuracy literature measures agreement rather than truth

Both figures for the one bedside test were produced by comparing it against standards that have never been shown to identify this condition. Any future accuracy work inherits this until somebody establishes an independent standard.

First-line conservative care rests on almost nothing, and the recommendation does not carry that with it

Measured directly at the index, the medical literature holds 8 randomized trials in total for this condition, and not one tests exercise, hands-on treatment or a physical therapy programme on its own. A patient who does not improve has not "failed physical therapy", because physical therapy was never established to work here. That framing changes what the next conversation is.

The Nuance

For a condition described in the medical literature since the 1970s, almost nothing is settled. No population prevalence. No validated test that can reliably clear the look-alikes. No controlled evidence for any conservative treatment. No threshold for meaningful change on any outcome measure. No agreed algorithm for escalating to surgery, and that last one is not an inference: a 2026 systematic review appraised 14 studies with three separate quality tools and found only a minority propose a structured algorithm at all (Nguyen 2026).

Surgery versus conservative care cannot be answered with numbers, because neither side has been measured against the other. Conservative success rate: never measured. Surgical success rate: reported only in uncontrolled case series.

The one thing the surgical literature establishes clearly is what these procedures do reliably. They cause numbness. Endoscopic neurotomy produced it in 21 of 21 patients and radiofrequency ablation in 48 of 55. Cutting the nerve root during upper neck surgery produced it at odds ratio 178.6 with no accompanying drop in occipital neuralgia. A patient weighing an irreversible procedure needs the numbness figure and the relief figure in the same sentence, and it is the numbness figure that rarely gets said out loud.

Sources

Melchior AG, 2025, Cephalalgia. Systematic review and meta-analysis, 15 clinic-based studies, N=579. No population-based studies exist; relative frequency 0.6 to 24.4 percent.
Deuel D, 2024, Current Pain and Headache Reports. Review. Conservative research specific to this condition is limited to very low-quality evidence for TENS.
Gul HL, 2017, Acta Neurologica Scandinavica. Placebo-controlled randomized trial, N=44 chronic migraine. Nerve block beat saline at months 1, 2 and 3.
Cohen SP, 2015, Pain. Multicentre double-blind randomized trial, N=81. Pulsed radiofrequency beat steroid injection at 6 weeks, sustained to 6 months.
Badhiwala JH, 2017, Spine. Meta-analysis, 8 studies, N=393. Occipital numbness odds ratio 178.6; occipital neuralgia odds ratio 1.44.
Samolsky Dekel BG, 2023, Pain and Therapy. Diagnostic accuracy study, N=163 consecutive patients.
Staudt MD, 2023, Neurosurgery. Congress of Neurological Surgeons guideline update. All 15 included studies are the weakest evidence class.
Saglam L, 2024, Journal of Anatomy. 41 cadavers dissected bilaterally. Four types, 18 subtypes, six compression points.
Huanmanop T, 2021, Folia Morphologica. 100 nerves from 50 cadaveric heads. 95 percent pierce semispinalis capitis.
Wang Y, 2018, Acupuncture Research. Three-arm randomized trial, N=62. The only randomized hands-on treatment arm in the world literature.
Ulutas S, 2026, Journal of Clinical Neuroscience. 2,219 patients, 2,250 nerve blocks, six countries.
Kollenburg L, 2024, Current Pain and Headache Reports. Systematic review of occipital nerve stimulation. Response 63 to 100 percent; lead migration or fracture 13 percent.
Huang Y, 2026, Frontiers in Pain Research. N=76. Numbness in 100 percent and 87.3 percent of the two treatment arms.
Nguyen AT, 2026, Plastic and Reconstructive Surgery Global Open. Systematic review, 14 studies. No standardized treatment algorithm exists.

Full 28-source evidence trail with study designs and effect sizes sits behind this page in the research record.