Headache with neck pain is one of the most common presentations in pain and headache clinics. Three conditions overlap in this group: cervicogenic headache, occipital neuralgia and migraine with neck pain. They are treated differently, and two nerves sit at the centre of the interventional options: the greater occipital nerve and the third occipital nerve. This article sets out what each nerve does, what blocking it can and cannot tell you, and how to move from clinical suspicion to a confirmed source and a targeted treatment.

1. What is cervicogenic headache?

Cervicogenic headache is pain referred to the head from a source in the cervical spine. The typical picture is a side-locked, unilateral, moderate, non-throbbing headache that starts in the neck or occiput and spreads forward to the frontal, temporal or orbital region.

ICHD-3 criteria

• B. Clinical or imaging evidence of a cervical disorder or lesion known to be able to cause headache.
• C. Evidence of causation, shown by at least two of the following:
  • the headache developed in temporal relation to the onset of the cervical disorder
  • the headache improved in parallel with improvement of the cervical disorder
  • cervical range of motion is reduced and the headache is made worse by provocative manoeuvres
  • the headache is abolished by diagnostic blockade of a cervical structure or its nerve supply
• D. Not better accounted for by another ICHD-3 diagnosis.

What “temporal relation” means in practice

The headache began at about the same time as the neck disorder, or soon after it. ICHD-3 gives no fixed time window, so the clinician judges whether the sequence is plausible.

• Meets the criterion: no headache before a road traffic collision, then occipital headache spreading forward that starts days to weeks later.
• Meets the criterion: headache that starts together with an acute upper cervical lesion, such as a fracture, disc herniation, infection or inflammatory arthritis flare.
• Does not meet it: long-standing episodic migraine, with spondylosis found later on imaging. The headache came first.

The temporal criterion is weak on its own, because degenerative change is common in people without headache. When the timing is vague, provocation on examination and the response to diagnostic blockade carry more weight. ICHD-3 allows both diagnoses when a pre-existing migraine or tension-type headache becomes significantly worse in close relation to a neck disorder.

2. Why neck pain is felt in the head: trigeminocervical convergence

Nociceptive afferents from the trigeminal nerve (mainly V1, which supplies the dura and the front of the head) and from the upper cervical nerves (C1–C3) synapse on the same second-order neurons in the trigeminocervical complex. This runs from the trigeminal nucleus caudalis down to the C2 dorsal horn, and to C3 in some descriptions.

Convergence has three practical consequences:

• Referral runs in both directions. The C2–3 joint, the atlanto-axial joint and the C2 root can produce frontal or orbital pain. Migraine and cluster headache commonly produce occipital and neck pain.
• The site of pain does not identify its source.
• An occipital nerve block can reduce pain in trigeminal territory. In a placebo-controlled fMRI study, a lidocaine GON block reduced trigeminal nociceptive activation at the C2 level. Relief often outlasts the local anaesthetic by days to weeks, which points to a central modulating effect, although the mechanism is inferred and not proven.

The same mechanism explains why a GON block is non-specific as a diagnostic test, a point developed in section 10.

3. Anatomy that matters

The greater and third occipital nerves compared

The C3 dorsal ramus has two medial branches

• Superficial medial branch (the TON). A thick nerve, about 1–1.5 mm, that crosses the C2–3 joint at a variable height. It gives articular branches to the joint, supplies semispinalis capitis, then becomes cutaneous. It communicates with the GON and the lesser occipital nerve.
• Deep medial branch. The “C3 medial branch” in the usual segmental sense. It runs around the waist of the C3 articular pillar and supplies the C3–4 joint together with the C4 medial branch. It is not cutaneous.
• Lateral branch. Supplies muscle.

Sources that say “the C3 medial branch is the TON” are using shorthand. The precise answer is the superficial medial branch.

Why the C2–3 joint is the exception

In most people the TON is the only nerve supply to the C2–3 joint. There is no C2 medial branch that supplies it in the way the lower medial branches supply their joints, because the medial branch of the C2 dorsal ramus continues as the GON and runs to the scalp. A small communicating branch from the C2 dorsal ramus sometimes contributes, and the deep C3 medial branch can run high enough to cross the lower part of the joint line. Both are inconstant.

Above C2–3 the rules change again. The lateral atlanto-axial joint is supplied by the C2 ventral ramus and the atlanto-occipital joint by the C1 ventral ramus. Neither has a medial branch that can be blocked or ablated.

The cervical numbering rule (and why it differs from the lumbar spine)

Below C2–3, each cervical facet joint is supplied by the two medial branches that share numbers with the vertebrae forming the joint. Each cervical nerve exits above its own vertebra and its medial branch wraps around the pillar of the same number, so one nerve lies just above the joint and one just below.

• C3–4 joint: deep C3 medial branch and C4 medial branch.
• C5–6 joint: C5 and C6 medial branches. Blocking C4 and C5 would not cover it.
• Lumbar contrast: the L4–5 joint is supplied by the L3 and L4 medial branches, because lumbar nerves exit below their own vertebra. The lumbar rule does not transfer to the neck.

Where each nerve sits on the articular pillar

C5 is the reference level. On a lateral view the medial branch crosses the centroid of the pillar, and on an AP view it lies at the lateral concavity (the waist). The nerves run obliquely, roughly parallel to the facet joints.

The TON and the deep C3 branch lie close together near the C3 superior articular process. A high deep C3 block can partly anaesthetise the C2–3 joint, and a low TON lesion can reach the deep C3 branch. Keep this in mind when interpreting blocks at C2–3 versus C3–4.

4. Which structures cause cervicogenic headache?

C2–3 is the most common source but not the only one. Govind and Bogduk investigated 166 patients with controlled blocks and found that the answer depends on whether headache or neck pain is the dominant complaint.

Two points are worth carrying into practice. Every study that identified C2–3 as the source was in patients with a history of trauma, usually whiplash. And occipital radiation predicts failure of standard C3–C6 medial branch radiofrequency treatment, which fits with the C2–3 joint being missed at those levels.

5. Clinical features and examination

Features that help most

• Unilateral headache that does not change sides.
• Pain that starts in the neck and spreads to the fronto-temporal or orbital region.
• Provocation by neck movement, sustained posture or pressure on the upper neck.
• Ipsilateral neck, shoulder or arm pain, and reduced cervical range of motion.
• A history of neck trauma before onset.

None of these is specific. Patients with block-confirmed cervicogenic headache have no distinctive clinical features, imaging findings overlap with those of people without headache, and tenderness over C2–3 has a positive likelihood ratio of only 2.1. The cervical flexion-rotation test shows reduced rotation on the symptomatic side and is widely used, although its reported accuracy comes mainly from the physiotherapy literature.

Can examination separate the atlanto-axial joint from C2–3?

Not reliably, but it can make one joint more likely and so decide which is blocked first.

A practical examination sequence

1. Seated, active movement. Test pure rotation at small angles (atlanto-axial), then extension followed by rotation to each side (facet joints, including C2–3).
2. Supine flexion-rotation. Fully flex the neck to lock the lower cervical spine, then rotate. Restriction or pain points to C1–2.
3. Prone palpation. Press the tip of the C1 transverse process, then apply posteroanterior pressure over the C2–3 pillar and the C3–4 pillar. A positive test reproduces the patient’s usual headache, not just local soreness.
4. Neurological examination. This should be normal. Any focal deficit points to another diagnosis.

6. C2–3 (TON) headache versus greater occipital neuralgia

These two are often confused because both cause occipital pain. They are separated mainly by pain quality and by what provokes the pain.

• Continuous occipital pain without dysaesthesia makes a cervical referral source (facet or atlanto-axial joint) more likely than occipital neuralgia.
• Lancinating pain with ipsilateral lacrimation or conjunctival injection points to the C2 root or ganglion, not the GON.
• Both can be present. C2–3 pain can coexist with GON entrapment or produce secondary GON sensitisation. Treat the confirmed joint source first, then reassess for residual neuralgic pain.
• Migraine tends to change sides. C2–3 headache and occipital neuralgia are side-locked.

7. Confirming the source: diagnostic blocks

Because clinical and imaging features are not valid diagnostic markers, controlled blocks are the only objective way to identify the source.

Principles of a valid diagnostic block

• Controlled. A single uncontrolled block has a significant false-positive rate. The Bogduk approach uses comparative blocks on separate occasions, lidocaine then bupivacaine, with relief that matches each drug’s duration. Most experts favour dual comparative blocks before TON radiofrequency treatment.
• Clear endpoint. Complete or near-complete relief of the headache, re-tested with the neck movements that normally provoke it. Lower thresholds, such as 50% relief, are used by some but increase false positives.
• Low volume. About 0.3–0.5 mL per target, so the block stays specific.
• No steroid. It adds nothing to the test.
• Documented. A headache diary for the hours after the block.

Which level first?

Start with the TON alone, because C2–3 has the highest pre-test probability. Blocking it in isolation keeps the result specific. Multilevel blocks show only that some joint in the blocked field is painful.

• If the TON block is negative and headache dominates: intra-articular C1–2 block next, then C3–4 (deep C3 and C4 medial branches).
• If the TON block is negative and neck pain dominates: C3–4 and the lower cervical medial branches.
• Combined blocks are reasonable when the picture clearly involves several levels, for example occipital headache with prominent mid-neck pain and tenderness at more than one level.
• A negative TON block does not exclude cervicogenic headache. It moves the search to other joints.

8. Third occipital nerve block: fluoroscopy or ultrasound

Both work. Fluoroscopy is the reference standard for diagnostic TON blocks and is required for radiofrequency treatment. Ultrasound is a validated alternative that shows the nerve itself and avoids radiation.

Fluoroscopic technique

1. Position. Prone, or lateral with the target side up.
2. True lateral view. Superimpose the left and right articular pillars so the C2–3 joint appears as a single line. A double contour means the C-arm is rotated.
3. Three targets. Just above the joint line, at the joint line and just below it, because the nerve’s height varies.
4. Needle. Advance until it touches bone on the lateral surface of the pillar at each point. Stay behind the anterior edge of the pillar.
5. AP check. The tip should lie on the lateral margin of the pillar and not medial to it.
6. Inject. Aspirate, consider live contrast to exclude vascular uptake, then give about 0.3–0.5 mL at each point.

Ultrasound technique

Figure 6. The third occipital nerve crosses the articulation of C2–C3

1. Find C2. With a high-frequency linear probe in long axis over the lateral neck, identify the mastoid, then the C1 transverse process, then C2. Alternatively, scan down from the occiput to the first bifid spinous process (C2) and slide laterally.
2. See the joint. The C2–3 joint appears as a bright convex “hill” between the two pillars. The TON is a small hypoechoic oval, about 2 mm, lying on it.
3. Check for vessels with colour Doppler.
4. Needle in plane, tip beside the nerve. Keep the volume small for a diagnostic block.

Safety. Unsteadiness is common after a TON block, so patients should not drive that day. Keep the needle on bone over the posterolateral pillar and aspirate before injecting, because the vertebral artery lies anterior to the pillars.

9. Third occipital nerve radiofrequency neurotomy

Selection and evidence

• Indication: cervicogenic headache from the C2–3 joint, confirmed by controlled TON blocks, after conservative treatment has failed.
• Outcome: Govind and colleagues performed comparative blocks in 120 patients with suspected third occipital headache. 49 were positive and proceeded to neurotomy. 88% had a successful outcome, with relief lasting around 297 days on average.
• Why selection matters: earlier negative studies selected patients on clinical criteria alone and lesioned C3–C6, which missed the C2–3 joint.
• Recurrence: the nerve regenerates. Repeat neurotomy is as effective as the first.

Technique

1. Position and view. Prone or lateral, with a true lateral view of the C2–3 joint.
2. Electrode parallel to the nerve. A conventional electrode coagulates around its shaft and very little beyond its tip, so it must lie along the nerve. Use a posterior or posterior-oblique approach, entering caudal to the target and angling cranially. The 2022 consensus does not support a purely perpendicular lateral approach for conventional electrodes.
3. Three overlapping lesions along the dorsolateral joint line: just above, at and just below it. Move the cannula less than one electrode width between lesions so no gap is left. Typical equipment is a 16–20 gauge cannula with a 10 mm active tip.
4. Depth. Keep the tip on bone and never anterior to the pillar, where the vertebral artery and the C3 ventral ramus lie. On the AP view the tip rests at the lateral concavity of the pillar, not medial to it.
5. Stimulation. Sensory testing at 50 Hz should produce suboccipital sensation below about 0.5 V. Motor testing at 2 Hz may cause a paraspinal twitch. Twitching in the arm, hand or face means the tip is too close to the ventral ramus and must be repositioned.
6. Lesion. Local anaesthetic through each cannula, then 80–85 °C for 60–90 seconds at each site.

Multi-tined (Trident) and cooled radiofrequency

• Trident cannula. It lesions in front of its tip, so it can be placed more end-on to the bone. It does not remove the need to cover the whole joint line. Cervical data are limited to a retrospective study showing shorter procedure and screening times, and there are no TON-specific comparative studies.
• Cautions. Because the lesion projects forward, the tip must stay on the dorsolateral pillar and behind its anterior edge. Always retract the tines before repositioning.
• Cooled radiofrequency is another way to produce a larger lesion and has a recent retrospective series in cervicogenic headache.

Side effects to discuss at consent

The TON has a cutaneous branch and supplies semispinalis capitis, so its ablation has side effects that are not seen after lower cervical medial branch treatment. The figures below come from older series and vary between reports.

10. The greater occipital nerve in management

What a GON block does and does not tell you

The GON does not supply the C2–3 joint or any other facet joint. A GON block works by reducing C2 input to the trigeminocervical complex, and it relieves migraine, cluster headache, occipital neuralgia and cervicogenic headache alike.

• A positive GON block is compatible with occipital neuralgia, but it does not exclude a facet, atlanto-axial, migraine or cluster source.
• It is not the gateway test for TON radiofrequency treatment. That requires controlled TON blocks.
• A negative GON block in a patient with a continuous, mechanical ache should prompt TON blocks.
• In controlled studies of migraine, tenderness over the GON did not predict response. Focal tenderness with a Tinel sign is better read as a cue to look for coexisting occipital neuralgia.

Where it is useful

Distal and proximal techniques

• Distal (landmark) block at the nuchal line. About one third of the way from the external occipital protuberance to the mastoid, just medial to the occipital artery. Typically 1.5–3 mL of lidocaine or bupivacaine.
• Proximal block at C2 under ultrasound. Identify the bifid C2 spinous process, slide laterally to OCI, and rotate the lateral end of the probe towards the C1 transverse process to see OCI in long axis. The GON is the hypoechoic oval in the plane between OCI and SSC. In a randomised trial in occipital neuralgia and cervicogenic headache, this approach reduced pain more than the landmark block at 30 minutes and at 4 weeks.
• This is an ultrasound procedure. The landmarks are soft tissue and cannot be seen on fluoroscopy.
• Depth matters. For a GON block the needle tip sits superficial to OCI. Injectate placed deep to OCI on the C2 lamina spreads to the C2 dorsal root ganglion, which makes it a different and less selective block. Larger volumes at C2 can also reach the TON.

Pulsed radiofrequency of the GON

For occipital neuralgia that responds to a low-volume block but relapses, pulsed radiofrequency of the GON has the best evidence. In a multicentre double-blind randomised trial of 81 patients, it reduced occipital pain more than steroid injection at 6 weeks, with benefit lasting to 6 months. Typical settings are 42 °C in 120-second cycles. Risks are minimal because the target is superficial.

11. When TON blocks are negative

Lateral atlanto-axial (C1–2) joint

• Suspect it when pain is rotation-dominant, sits behind the mastoid, and there is tenderness over the C1 transverse process. Clinical selection alone has a positive predictive value of only about 47%.
• Confirm with a fluoroscopic intra-articular block by a posterior approach, aimed at the lateral third of the joint to stay medial to the vertebral artery and lateral to the C2 ganglion and dura. Use an arthrogram and less than 0.5 mL of injectate.
• Ultrasound alone is not validated for this joint. It is useful alongside fluoroscopy to map the vertebral artery.
• Treatment: intra-articular steroid gives short-term relief in observational series. There is no medial branch to ablate. C1–2 fusion is reserved for refractory osteoarthritis.

C2 dorsal root ganglion

• What it covers: the whole C2 segment, including the GON, part of the lesser occipital nerve and the atlanto-axial joint. It is therefore not equivalent to a GON block and has low selectivity for any single structure.
• Consider it when a GON block gives partial or short-lived relief, when pain spans the C2 dermatome, or when there are features of C2 neuralgia such as lacrimation or conjunctival injection.
• Guidance: fluoroscopy or CT with contrast is the best-validated approach. An ultrasound approach deep to OCI on the C2 lamina is increasingly used by experienced operators.
• Pulsed radiofrequency: in a retrospective series of 45 patients, 40% had at least 50% relief at 6 months. A clearly positive diagnostic block was the strongest predictor of success.
• Risks: vertebral artery injection, dural puncture and intrathecal spread. Use low volumes and contrast.

Atlanto-occipital joint

This is an uncommon source, provoked more by nodding and lateral bending than by rotation. Injection is higher risk than at C1–2 because the vertebral artery crosses the joint, and there is no denervation target afterwards. It is usually considered only after the C2–3 and C1–2 joints have been tested.

12. A practical pathway

1. Exclude red flags and image where indicated.
2. Phenotype the pain. Mechanical and referred (continuous ache, worse with movement, pillar tenderness, trauma history), or neuropathic (paroxysms, allodynia, Tinel sign over the GON), or migrainous.
3. Mechanical phenotype: controlled TON blocks first. If positive, TON radiofrequency neurotomy.
4. TON blocks negative: C1–2 intra-articular block if headache dominates, or C3–4 and lower medial branches if neck pain dominates.
5. Neuropathic phenotype: low-volume GON block. If relief is clear but short-lived, GON pulsed radiofrequency. If relief is partial or the pain extends beyond GON territory, consider a C2 dorsal root ganglion block.
6. Mixed or unclear: start with the TON block if C2–3 tenderness or provocation is present, because a positive GON block cannot localise a cervical source.
7. Throughout: treat coexisting migraine on its own merits, and combine interventions with rehabilitation.

Key sources

• Bogduk N, Govind J. Cervicogenic headache: an assessment of the evidence on clinical diagnosis, invasive tests, and treatment. Lancet Neurology. 2009.
• 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. Regional Anesthesia and Pain Medicine. 2022.
• Govind J, Bogduk N. Sources of cervicogenic headache among the upper cervical synovial joints. Pain Medicine. 2022.
• Lord SM, et al. Percutaneous radio-frequency neurotomy for chronic cervical zygapophyseal-joint pain. New England Journal of Medicine. 1996.
• Ng A, Wang D. Cervical facet injections in the management of cervicogenic headaches. Current Pain and Headache Reports. 2015.
• Barmherzig R, Kingston W. Occipital neuralgia and cervicogenic headache: diagnosis and management. Current Neurology and Neuroscience Reports. 2019.
• Mathew PG, Sailesh S. A pain in the neck: occipital neuralgia vs cervicogenic headache vs migrainous cervicalgia. Current Pain and Headache Reports. 2025.
• Fabry A, Nedunchelian M, Stacoffe N, et al. Review of craniofacial pain syndromes involving the greater occipital nerve. Neuroradiology. 2024.
• Stern JI, Chiang CC, Kissoon NR, Robertson CE. Narrative review of peripheral nerve blocks for the management of headache. Headache. 2022.
• Narouze S. Occipital neuralgia diagnosis and treatment: the role of ultrasound. Headache. 2016.
• Eichenberger U, et al. Sonographic visualization and ultrasound-guided block of the third occipital nerve. Anesthesiology. 2006.
• Kim HS, Shin KJ, O J, et al. Stereotactic topography of the greater and third occipital nerves and its clinical implication. Scientific Reports. 2018.
• Aprill C, et al. Occipital headaches stemming from the lateral atlanto-axial (C1–2) joint. Cephalalgia. 2002.
• Abd-Elsayed A, Hashimoto N, Reilly M, et al. Cooled radiofrequency ablation of cervical nerves for treating cervicogenic headache. Current Pain and Headache Reports. 2026.

This article is written for healthcare professionals and is for educational purposes. It does not replace clinical judgement, formal training in interventional procedures or local guidelines.