When the Headache Starts in Your Neck: Occipital Neuralgia, Cervicogenic Headache, and the Test That Comes Closest to an Answer

What this video covers

• How upper cervical nerves converge with trigeminal fibers and refer pain to the eye, temple and forehead
• The workup that separates occipital neuralgia, cervicogenic headache and migraine
• Why a diagnostic block answers a question before it treats anything
• What the randomized data on occipital nerve blockade actually shows — and what it does not: the trial followed patients only two weeks, so nothing is known from it about durable relief. A local anesthetic block lasts hours to a day or two and is mainly a diagnostic test; adding steroid may extend that to weeks in some patients, on weaker evidence
• Third occipital nerve radiofrequency neurotomy: 43 of 49 patients had complete relief lasting at least 90 days, median 297 days, in a single-center uncontrolled series by the group that developed the technique — and only in patients whose diagnosis was first confirmed by controlled comparative diagnostic blocks
• Material risks and expected effects: unsteadiness and ataxia in most patients and scalp numbness in about half after third occipital nerve neurotomy, painful post-neurotomy neuralgia in a minority, bleeding, infection, vasovagal reaction, allergic reaction, intravascular injection with local anesthetic toxicity, no relief at all, and with steroid, higher blood sugar plus hair loss, skin thinning and depigmentation at the injection site
• MEDICAL DISCLAIMER: This content is for educational purposes only and is not medical advice. It does not substitute for professional diagnosis or treatment. Always consult a licensed healthcare provider regarding your condition. Viewing this video does not establish a doctor-patient relationship.

It begins at the base of your skull, on one side, always the same side. A deep, boring ache under the occiput that climbs over the back of your head, wraps the temple, and settles behind the eye. Turning to check a blind spot sets it off. Sleeping wrong sets it off for two days. You have been told it is migraine — triptans, preventives, a neurologist, a normal brain MRI, a shrug. This article explains a different possibility: that head pain which starts in the neck is not a variant of migraine, and is diagnosed a different way.

The wiring that lets a neck problem feel like a headache

The anatomy explains this, and it is worth understanding before anyone offers you a fourth preventive.

The sensory fibers of the first three cervical nerves — C1, C2, and C3 — enter the upper cervical dorsal horn, the first relay station in the spinal cord at the top of the neck. That dorsal horn is anatomically continuous with the descending trigeminal nucleus, the relay that carries sensation from your face, your eye, and your forehead. Cervical and trigeminal fibers converge onto shared second-order neurons in what is called the trigeminocervical nucleus.

Two very different territories, in other words, report to the same switchboard. That convergence — supported by animal electrophysiology and human referral mapping — best explains why an irritated structure at the back of your neck produces pain felt at the orbit, the temple, and the vertex. It is the mechanism, stated at its actual strength: a well-supported explanation, not a bedside measurement anyone performs on you.

The specific structures that feed that line

Four named structures sit upstream of that shared nucleus, and any of them can be the generator:

• The greater occipital nerve, arising from the C2 dorsal ramus.
• The lesser occipital nerve, from the cervical plexus.
• The third occipital nerve, the superficial medial branch of the C3 dorsal ramus. It crosses the lateral C2-3 facet joint and supplies it.
• The C2-3 zygapophysial (facet) joint itself — a recognized source of headache after whiplash.

They sit close to one another and they all report through the same convergence. Symptoms alone cannot tell them apart. That single sentence is the reason this problem is so often miscategorized.

Why the pain lands behind your eye when the problem is under your skull

Patients describe this pattern so consistently that it should be a clue rather than a curiosity: pain that ignites suboccipitally, travels over the vertex, and pools behind the orbit on the same side, every time.

That is convergence doing exactly what convergence does. Your brain receives a signal on a pathway shared by the neck and the face, and it has no reliable way to know which end of the pathway the signal started at. So it assigns the pain to the face and the eye — territory that feels far more “headache” than “neck.”

The movement pattern is the second clue. Because the generator is a joint or a nerve that lives in the neck, the pain answers to neck mechanics. Rotation toward the painful side can reproduce it. A blind-spot check sets it off. A night in a bad position sets it off for two days. That mechanical behavior is precisely what the standard headache workup does not test for.

The side-locking is worth noticing too. Cervicogenic headache in particular presents as a side-locked ache, because the structure generating it sits on one side. It is a reason to examine the upper neck — not, on its own, a diagnosis. Symptoms alone still cannot tell these conditions apart.

What the standard headache pathway misses

None of this is a failure of any individual physician. It is a routing problem.

Headache is coded as a brain problem, so it is routed to a brain algorithm — and that algorithm has no step that says examine the upper neck. In a fifteen-minute headache visit, the things that go undone are predictable: nobody palpates the C2-3 joint. Nobody tests whether rotation to the painful side reproduces the headache. Nobody asks about a rear-end collision or a fall from years earlier. A brain MRI is ordered, comes back normal, and normal is treated as an answer when it is only the absence of one kind of answer.

Medication management has its place in headache care. What it cannot do is identify which structure is generating the pain. That is a different question, and it takes an examination and a procedure to resolve — not a longer medication list.

First, the dangerous causes have to be excluded

Before any of the above matters, catastrophe has to be ruled out. This part is not optional and it is not a pain-clinic problem.

A new, abrupt, or thunderclap headache requires urgent evaluation, not an injection. So does headache accompanied by any of the following:

• neurologic deficit
• fever
• jaw claudication (pain in the jaw with chewing)
• vision loss
• scalp tenderness in an older patient
• signs of spinal cord compression

Carotid or vertebral artery dissection, intracranial mass or hemorrhage, cervical myelopathy, and giant cell arteritis are evaluated urgently — they are not injected. Only when those branches are dark does the benign differential begin.

Occipital neuralgia, cervicogenic headache, and migraine — how they differ

Occipital neuralgia is a nerve problem. The pain is lancinating and electric, following the distribution of the greater, lesser, or third occipital nerve — most commonly the greater. There is often a tender point where the nerve pierces the fascia, and frequently altered sensation in that patch of scalp.

Cervicogenic headache is a referral problem — pain generated by a cervical structure and felt in the head. Most often that structure is the C2-3 joint, but it can be the disc, the deep muscle, or the occipital nerve itself. It presents as a steadier, side-locked ache provoked by neck position and movement, with restricted rotation on examination.

Migraine has its own signature: nausea, photophobia, phonophobia, throbbing quality, family history.

They overlap, because they share that nucleus. History and examination narrow the field without closing it. The test that comes closest to an answer is anesthetic, not radiographic — and only when it is repeated under control.

The block as a question, not as a treatment

Here is the step that gets skipped almost everywhere.

A diagnostic block places a small volume of local anesthetic onto one named target under image guidance — ultrasound for the occipital nerves, fluoroscopy for the third occipital nerve and the cervical medial branches — and then measures what your headache does over that anesthetic’s known duration. It is asking a question. It is not, by itself, the treatment.

And one positive block is not an answer. Barnsley and Bogduk showed in the Clinical Journal of Pain in 1993 that a single uncontrolled cervical block is falsely positive roughly twenty-seven percent of the time. That is precisely why one block is never enough — and also why the block still has to be done. Nothing else interrogates the source directly.

The fix is a controlled comparative block: the block is repeated, weeks later, with an anesthetic agent of a different duration. What you are looking for is concordant relief across two blocks that behaves the way each agent should — not enthusiasm after one.

The multispecialty consensus guidelines published by Hurley and colleagues in Regional Anesthesia and Pain Medicine in 2022, on cervical facet joint pain, favor medial branch blocks over intra-articular injection as the more predictive test. (Established guideline position — but note that this guidance addresses cervical facet joint pain, not headache specifically.) The procedural detail of how these blocks are targeted and staged is set out on the severe head and neck pain treatment page.

One more point of honesty about the medication in that syringe. A plain local anesthetic block lasts hours to a day or two; its value is informational. Adding corticosteroid may extend relief to weeks in some patients, on weaker evidence, and that use is off-label — no injected medication carries FDA approval for occipital neuralgia or cervicogenic headache. Do not start, stop, or change any medication without consulting your physician.

What the evidence actually shows — at its true weight

This is the section that usually gets sanded down. It should not be.

The randomized data. Naja and colleagues published a double-blind, saline-controlled randomized trial in cervicogenic headache in Pain Practice in 2006, with forty-seven patients analyzed. Nerve-stimulator-guided occipital blockade reduced pain scores roughly fifty percent from baseline and cut analgesic consumption significantly more than saline. (Established finding — but from a single small randomized, saline-controlled trial: short-term and imprecise.) Three qualifications travel with it, and each one matters:

1. Follow-up was only two weeks. That trial tells you nothing about durable relief.
2. The injectate was bupivacaine, lidocaine, clonidine and fentanyl. The clonidine and the fentanyl are off-label for peripheral nerve blockade.
3. Because of that mixture, this trial does not validate the plain local anesthetic block, or the anesthetic-plus-steroid block, that you would actually be offered.

It is the best randomized signal that exists here. It is not a result anyone should expect to be handed. Individual results vary.

The neurotomy data. If comparative blocks confirm a third occipital nerve source, radiofrequency neurotomy — heating the nerve to interrupt the signal — becomes a reasonable discussion. Govind, King, Bailey and Bogduk reported in the Journal of Neurology, Neurosurgery and Psychiatry in 2003 that forty-three of forty-nine patients met a hard endpoint: complete relief for at least ninety days, normal daily activities restored, and no headache medication at all — with a median duration of two hundred ninety-seven days.

Read the design before you read the number. That is a single-center, open-label, uncontrolled series from the group that developed the technique, in patients selected by controlled comparative blocks. There is no sham-controlled trial of this procedure. The figure should not be inflated, and looser patient selection or different technique would be expected to produce lower numbers. (Low-certainty evidence — uncontrolled series.)

And the relief is a window, not a cure. The nerve regenerates. In that same series, fourteen patients had the neurotomy repeated when the nerve grew back; twelve regained relief, with a median of two hundred seventeen days. The honest shape of this is a repeatable window — not a one-time cure, and not a single throw of the dice. Individual results vary.

What about nutrition and light therapy? Ketogenic nutrition, specialized pro-resolving mediators, and red and near-infrared photobiomodulation remain adjunctive and emerging for this condition, with largely preclinical mechanisms and no controlled human outcome data in occipital neuralgia or cervicogenic headache. (Investigational.) Support, never the answer.

The risks, said out loud

After third occipital nerve neurotomy, some effects are expected rather than rare. Unsteadiness and ataxia occur in most patients. Scalp numbness occurs in about half. Painful post-neurotomy neuralgia occurs in a minority.

Beyond that: bleeding, infection, vasovagal reaction, allergic reaction, intravascular injection with local anesthetic toxicity, and no relief at all. When corticosteroid is used, add elevated blood sugar plus hair loss, skin thinning, or depigmentation at the injection site.

Relief is durable, not permanent. Any description of this procedure that omits the regeneration is incomplete, and the expected effects above should be discussed in full before consent, not discovered afterward.

An honest picture of a good outcome

What follows is a composite — a picture assembled from many patients with this condition, not one person’s chart.

Consider someone rear-ended, recovered from the obvious injuries, left months later with right-sided headache running from occiput to eye. Four years of migraine treatment. Normal imaging. On examination, rotation to the right is limited and reproduces the headache, with tenderness over the C2-3 joint. A fluoroscopically guided third occipital nerve block abolishes the headache for the anesthetic’s duration. A comparative block weeks later, with a different agent, does the same. Only then is neurotomy offered.

The first weeks are genuinely unpleasant. The world tilts when they walk, a patch of scalp feels like leather, and someone else drives. The neck still has to be rehabilitated — deep cervical flexor endurance, scapular control, postural work. Sleep, glycemic control, and inflammatory load are worth attending to in anyone carrying chronic pain, but there is no data to say they drive this particular headache, so no such claim is made here. Somewhere in the third month, the headache stops being the first thing they think about. Not gone. Not first. Individual results vary.

What this means for your evaluation

If your headache is one-sided, starts below the skull, and moves with your neck, the request to make is specific: ask for the differential you were never given.

1. Rule out the dangerous causes first. New, abrupt, or thunderclap headache, neurologic deficit, fever, jaw claudication, vision loss, scalp tenderness in an older patient, or cord compression signs are urgent evaluations, not procedures.
2. Insist the upper neck is actually examined. Palpation of the C2-3 joint, rotation testing, and a history that includes old collisions and falls.
3. Ask whether a diagnostic block is appropriate — and understand it as a question being asked, with a measured answer over a known duration.
4. Insist that any ablation be preceded by two concordant blocks, not one. The twenty-seven percent false-positive rate of a single uncontrolled block is the whole reason.
5. Rehabilitate the neck regardless. No procedure substitutes for deep cervical flexor endurance, scapular control, and postural work.

Frequently asked questions

Key takeaways

• Upper cervical nerves converge with trigeminal fibers in the trigeminocervical nucleus, which is why an irritated structure under the skull can be felt behind the eye.
• Four structures can generate it — the greater, lesser, and third occipital nerves and the C2-3 facet joint — and symptoms alone cannot tell them apart.
• Dangerous causes such as arterial dissection, intracranial pathology, myelopathy, and giant cell arteritis are excluded urgently before any procedure is considered.
• A single positive block is falsely positive roughly twenty-seven percent of the time; the standard is concordant relief on two controlled comparative blocks.
• The evidence is honest but limited: one small randomized trial with two-week follow-up and a non-standard injectate, and one uncontrolled open-label neurotomy series with no sham control.

Medically reviewed by Dr. Gurpreet Singh Padda, MD, MBA, MHP — Board Certified in Anesthesiology, Pain Medicine, Interventional Pain Management, Addiction Medicine, and Obesity Medicine. Last reviewed July 2026.

This article is educational and is not a substitute for evaluation, diagnosis, or treatment by a physician. Individual results vary. Do not start, stop, or change any medication without consulting your physician. To have one-sided head and neck pain properly worked up before any ablation is scheduled, call (314) 481-5000 or text (314) 886-5902.

References

1. Govind J, King W, Bailey B, Bogduk N. Radiofrequency neurotomy for the treatment of third occipital headache. J Neurol Neurosurg Psychiatry. 2003;74(1):88-93. PMID 12486273. PubMed
2. Naja ZM, El-Rajab M, Al-Tannir MA, Ziade FM, Tawfik OM. Occipital nerve blockade for cervicogenic headache: a double-blind randomized controlled clinical trial. Pain Pract. 2006;6(2):89-95. PMID 17309715. PubMed
3. Hurley RW, Adams MCB, Barad M, et al. Consensus practice guidelines on interventions for cervical spine (facet) joint pain from a multispecialty international working group. Reg Anesth Pain Med. 2022;47(1):3-59. PMID 34764220. PubMed

Headache that will not let up?

A severe headache that is not an emergency still should not wait weeks for evaluation. The Padda Institute holds same-day and emergency visits at our Woodson Road office — no referral needed, and you do not need to be an existing patient. Call (314) 481-5000 or text (314) 886-5902.

Dr. Gurpreet Singh Padda, MD, MBA, MHP