Cluster Headache
Not a bad migraine — a different disorder, with a treatment most sufferers are never offered
What it is
Cluster headache is a primary headache disorder and one of the most severe pains in medicine. It is nothing like a migraine to experience. Attacks are strictly one-sided, centred in or around the eye, come on fast, peak within minutes, and last 15 to 180 minutes untreated. They come with autonomic features on the same side as the pain — a watering red eye, a drooping or swollen eyelid, a blocked or running nostril, a sweating forehead — and with something almost unique to this disorder: agitation. Where a person with migraine lies still in a dark room because movement makes it worse, a person in a cluster attack paces, rocks, or bangs their head, because stillness is unbearable. The pattern is the other signature. Attacks arrive in bouts (clusters) lasting weeks to months, often at the same times each day and often at the same time of year, with remission in between — that is episodic cluster headache. Chronic cluster headache runs without meaningful remission. It is uncommon, affecting roughly one in a thousand people, and it is more common in men, which is the reverse of migraine.
Why it matters
Cluster headache is defined by two failures, and both are addressable. The first is diagnostic: it is routinely mistaken for migraine, for sinusitis (because of the blocked nose and the eye), or for dental problems — people have had teeth extracted for it — and the average delay to diagnosis is measured in years, not months. Every one of those years is spent in a pain severe enough to have earned the disorder its grim nickname, and severe enough that suicidal thoughts during attacks are a documented feature rather than a figure of speech. The second failure is that the single best acute treatment is a gas. High-flow oxygen aborts most attacks within fifteen minutes, has essentially no side effects, and is under-prescribed almost everywhere — partly because it is not a tablet, partly because home oxygen requires arranging, and partly because clinicians who have not seen the disorder do not think of it. This page exists mainly to close those two gaps: to name the pattern so it can be recognised, and to say plainly that oxygen works and should be asked for.
What BioSignal knows about treating this
What works for Cluster Headache
BioSignal’s clinical summary, most important first.
- High-flow oxygen via a non-rebreather mask — the best-evidenced acute treatment, and the most under-prescribed
- Subcutaneous sumatriptan — fast and effective for aborting attacks; the injection, not the tablet, because tablets are too slow for a 45-minute attack
- Verapamil — the mainstay of prevention during a bout; requires ECG monitoring because the doses used are high
- Galcanezumab — reduces attack frequency in episodic cluster headache, on modest and imprecise trial evidence
- Greater occipital nerve blocks and short steroid courses — used as a bridge while verapamil takes effect
- Stopping smoking, and avoiding alcohol during a bout
- Specialist referral — this is a disorder where a headache specialist changes outcomes, and where most people wait years to reach one
Signal Records relevant to this condition
Interventions and contributing factors — some of these records describe a cause rather than a cure. The rating shown is BioSignal’s confidence in that Signal Record, not a claim about how well it treats this condition. Open any of them for the full evidence.
- Smoking CessationHigh confidence
One of the highest-value interventions in medicine — roughly a decade of life, and about 90% of the excess risk avoided if you stop before 40. The treatments are well evidenced and unequal: varenicline, nicotine e-cigarettes and cytisine roughly double quit rates, combination NRT is close behind, and a single patch or willpower alone is considerably weaker. Nicotine replacement is not as dangerous as smoking, and believing otherwise keeps people smoking.
- AlcoholHigh confidence
A Group 1 carcinogen with dose-dependent harms. The heart-protection belief does not survive the methods designed to test it. Cutting back helps, even without quitting.
New to this? Read these first
How this usually unfolds
- Recognize risk factors
- Get diagnosed
- Track key biomarkers
- Lifestyle first
- Evidence-based treatment
- Long-term monitoring
Who is at risk
- Male sex — the reverse of migraine's pattern
- Smoking — strongly overrepresented among people with cluster headache
- Family history
- Age 20-40 at onset, typically
- Alcohol — a reliable trigger DURING a bout, and typically harmless between bouts, which is itself diagnostic
How it's diagnosed
Cluster headache is diagnosed clinically on the pattern, using ICHD-3 criteria, and the pattern is distinctive enough that the diagnosis is usually available to anyone who asks the right questions. The combination that identifies it: severe, strictly one-sided pain around the eye or temple; duration of 15 to 180 minutes untreated; autonomic features on the same side; restlessness or agitation during the attack; and a striking frequency, from one attack every other day up to eight a day. The circadian and circannual rhythm — attacks at the same hour, bouts at the same season — is close to unique among headache disorders and is worth volunteering to a doctor, because it is the detail that most often flips the diagnosis. Brain imaging with MRI is recommended at least once, not because cluster headache shows on it, but because a small number of structural lesions can produce an identical picture (a 'symptomatic' cluster-like headache), and that possibility is worth excluding once. Beyond that, there is no test: the diagnosis is the story, which is exactly why telling it accurately matters more here than in most conditions.
Key biomarkers
Biomarker pages for this condition are on the roadmap.
Lifestyle
Explore this condition across BioSignal
Related Foundations
Related Signal Records
Related body systems
Related conditions
Frequently asked questions
Is cluster headache just a very bad migraine?
No — it is a different disorder, and treating it as migraine is why people lose years. The differences are not subtle once you know them. Cluster attacks are strictly one-sided and centred on the eye; they last 15 to 180 minutes rather than 4 to 72 hours; they come with autonomic features on the same side — a red watering eye, a drooping lid, a blocked or streaming nostril, a sweating forehead; and they can strike up to eight times a day. The single most telling difference is behavioural: migraine makes people lie still in the dark because movement worsens it, while cluster headache makes people restless and agitated — pacing, rocking, unable to stay in a chair. If someone cannot keep still during an attack, that is not a migraine. And the rhythm is close to unique: attacks at the same hour, bouts at the same season. This matters because the treatments barely overlap. Migraine treatments do little here, while the things that work for cluster headache — high-flow oxygen, injected sumatriptan, verapamil — are not standard migraine care. Being told you have migraine when you have cluster headache is not a labelling error; it is years of the wrong treatment for one of the worst pains there is.
Does oxygen really work?
Yes, and it is the most under-used effective treatment in this entire field. In a randomised trial, patients treated attacks with inhaled high-flow oxygen or with air: 78% were pain-free at 15 minutes with oxygen, compared with 20% with air, and there were no important adverse events. That is a large effect, achieved quickly, with a treatment that has essentially no toxicity, no interactions, no dose ceiling and no rebound. It is also a gas, which is the whole problem — it requires a cylinder, a non-rebreather mask and a prescription, none of which is as easy as handing over a packet, and many clinicians who rarely see this disorder simply do not think of it. If you have cluster headache and have never been offered oxygen, that is worth raising explicitly and by name. Two practical points that come with it: the delivery matters (high flow, through a non-rebreather mask — a nasal cannula does not do this job), and oxygen is an acute treatment for the attack in front of you, not a preventive. It does not reduce how many attacks you get; it ends them.
What stops the attacks coming?
Prevention during a bout, and verapamil is the mainstay. It is used at doses considerably higher than in cardiology, which is why it needs ECG monitoring — the risk being conduction problems in the heart — and it is not a drug to escalate on your own. Because it takes time to work, it is common to bridge with a short course of steroids or a greater occipital nerve block, which can buy relief while the verapamil comes up to dose. Galcanezumab, a CGRP monoclonal antibody, is the newer option, and BioSignal will report its evidence as it stands rather than as it is marketed: in the randomised trial, weekly attacks fell by 8.7 in the galcanezumab group versus 5.2 with placebo — a difference of 3.5 attacks per week, with a confidence interval running from 0.2 to 6.7 and a p-value of 0.04. Note two things in that. The placebo group improved substantially, which is what bouts do on their own. And the trial stopped recruiting before reaching its planned size because too few people qualified, leaving the estimate imprecise. So galcanezumab is a real but modest option on modest evidence, in episodic cluster headache specifically. None of this is a plan you can build alone — dose, monitoring and sequence are specialist decisions.
Why does it get mistaken for sinus problems or toothache?
Because of where it is and what it does to the face. The pain sits behind or around the eye and in the upper jaw and temple, and the autonomic features — a blocked or running nostril, a watering eye, facial sweating — look exactly like sinus disease to someone who is not thinking about headache disorders. So people are treated for sinusitis, given antibiotics that cannot help, sent for sinus imaging, and sometimes referred to dentists, where teeth have been drilled and extracted for a pain that was never coming from them. The tell is the pattern, not the location: sinusitis does not produce excruciating attacks that start and stop within an hour, up to eight times a day, at the same time each day, in bouts separated by months of nothing. If you have been treated repeatedly for sinus infection or dental pain without lasting benefit, and your pain arrives in discrete severe attacks with a rhythm, that history is worth putting in front of a doctor as a whole — because it is the shape of it, not any single feature, that identifies this.
How bad is it, and is it dangerous?
It is among the most severe pains described in medicine, and BioSignal is not going to soften that — but severity and danger are different things, and both parts of that deserve saying. Cluster headache does not damage the brain, and it is not a sign of something growing in your head; it is a primary headache disorder, and the attack that feels catastrophic is not causing harm. What it does do is cause suffering intense enough that suicidal thoughts during attacks are a recognised and documented feature of the condition, which is the honest reason its old nickname exists. If that is you, it is not a sign of weakness and it is not a reason to wait: it is a reason to be seen urgently and to be treated properly, because this disorder is treatable and most people who feel that way are undertreated rather than untreatable. If you are having thoughts of harming yourself, please contact emergency services or a crisis line now — that part is not something to manage alone or to read about. The pain is the emergency's cause, and it has answers.
When should a headache like this be checked urgently?
The first time, and whenever it changes. A first-ever severe one-sided headache with a red, watering, drooping eye needs assessment rather than self-diagnosis — a small number of structural problems, including some involving the arteries in the neck and the base of the skull, can produce a picture indistinguishable from cluster headache, which is why guidance is to image the brain at least once even when the pattern is classic. Seek urgent care for a headache that is sudden and reaches maximum intensity within a minute, for headache with weakness, numbness, difficulty speaking, or a drooping face, for headache with fever and neck stiffness, for new visual loss, or for a headache after head injury. Also seek review if your established pattern changes — attacks that switch sides, that stop responding to what worked, or that lose their rhythm are worth re-examining rather than absorbing. BioSignal does not triage any of these, and the reason to list them is specific: cluster headache is severe enough that people learn to endure, and endurance is exactly what hides a change worth noticing.
Evidence summary
Cluster headache is a primary headache disorder distinct from migraine, characterised by strictly unilateral severe periorbital pain lasting 15-180 minutes untreated, ipsilateral cranial autonomic features, restlessness or agitation during attacks (in contrast to migraine's motion avoidance), attack frequency up to eight daily, and a circadian and circannual rhythm that is close to diagnostic. It affects roughly 0.1% of people and, unlike migraine, is more common in men; smoking is strongly overrepresented, and alcohol reliably triggers attacks during a bout while typically not between bouts. Diagnosis is clinical against ICHD-3 criteria; MRI is recommended at least once to exclude the rare structural mimics, not to establish the diagnosis. Diagnostic delay is a defining problem, with misattribution to migraine, sinusitis and dental disease common. High-flow oxygen is the best-evidenced acute treatment and the most under-prescribed: in a randomised trial, 78% were pain-free at 15 minutes with high-flow oxygen versus 20% with air, with no important adverse events — a large effect from a treatment with essentially no toxicity, whose main barrier is logistical rather than clinical. Subcutaneous sumatriptan is effective for aborting attacks; oral formulations are too slow for an attack that peaks in minutes. Verapamil is the mainstay of preventive treatment during a bout, at doses requiring ECG monitoring for conduction effects, commonly bridged with short steroid courses or greater occipital nerve blocks. Galcanezumab reduced weekly attack frequency in episodic cluster headache (8.7 vs 5.2 attacks, difference 3.5 per week, 95% CI 0.2-6.7, P=0.04; ≥50% reduction in 71% vs 53%), but the trial halted recruitment below its planned sample size because too few volunteers qualified, leaving an imprecise estimate against a substantially improving placebo arm — a real but modest option on modest evidence. Current EAN guidance and recent reviews inform management. Suicidal ideation during attacks is a documented feature of the disorder and reflects undertreated pain rather than a psychiatric primary. This page covers cluster headache; migraine and tension-type headache are separate published Conditions, sinusitis is a separate Condition, and medication-overuse and cervicogenic headache remain unpublished.
References & sources
- Headache Classification Committee of the International Headache Society (IHS). The International Classification of Headache Disorders, 3rd edition. Cephalalgia 2018;38(1):1-211 (PMID 29368949; DOI 10.1177/0333102417738202)
- May A, Evers S, Goadsby PJ, et al. European Academy of Neurology guidelines on the treatment of cluster headache. Eur J Neurol 2023;30(10):2955-2979 (PMID 37515405; DOI 10.1111/ene.15956)
- Petersen AS, Lund N, Goadsby PJ, et al. Recent advances in diagnosing, managing, and understanding the pathophysiology of cluster headache. Lancet Neurol 2024;23(7):712-724 (PMID 38876749; DOI 10.1016/S1474-4422(24)00143-1)
- Cohen AS, Burns B, Goadsby PJ. High-flow oxygen for treatment of cluster headache: a randomized trial. JAMA 2009;302(22):2451-2457 (PMID 19996400; DOI 10.1001/jama.2009.1855)
- Goadsby PJ, Dodick DW, Leone M, et al. Trial of galcanezumab in prevention of episodic cluster headache. N Engl J Med 2019;381(2):132-141 (PMID 31291515; DOI 10.1056/NEJMoa1813440)
- Leone M, D'Amico D, Frediani F, et al. Verapamil in the prophylaxis of episodic cluster headache: a double-blind study versus placebo. Neurology 2000;54(6):1382-1385 (PMID 10746617; DOI 10.1212/wnl.54.6.1382)
Educational information — not medical advice
Is this condition page clear, accurate, and useful? Your feedback shapes what we review next.
Give feedback