Types of Headaches: What the Primary and Secondary Split Means for Diagnosis

Every headache is either a disease in its own right or a symptom of a separate one, and that split, primary versus secondary, organizes the entire official classification. A primary headache is the disorder itself: nothing else is producing it, and the pain plus its accompanying symptoms are the whole condition. A secondary headache is pain generated by something separate, whether that is a viral fever, a concussion, a course of painkillers taken too often, or a torn artery in the neck. Migraine, tension-type headache, and cluster headache sit on the primary side. Headache from sinusitis, from a bleed, from caffeine withdrawal, or from a spinal tap sits on the secondary side.

This article explains a medical topic in general terms. It is not medical advice, it is not reviewed by a clinician, and it cannot account for your individual case. Decisions about your own diagnosis or treatment belong with your doctor.

That distinction does more work than any description of pain quality. Two people can both report a dull band of pressure across the forehead, and one has a common tension-type headache while the other has raised pressure inside the skull. What separates them is the pattern around the pain: how it started, how it has behaved over the weeks since, what the examination shows, and whether anything else in the picture still needs explaining. Below: how the classification is built and who maintains it, the named types inside each category, how a clinician narrows the field, and which presentations the published guidance treats as emergencies.

What counts as a headache type, and who decides?

Headache types are defined by the International Classification of Headache Disorders, now in its third edition and usually shortened to ICHD-3, published by the International Headache Society. It is the reference that neurologists, emergency physicians, and headache researchers work from, and it is why a migraine diagnosis in Osaka means the same thing as a migraine diagnosis in Toronto.

The structure is worth understanding because it explains a lot of otherwise confusing medical language. ICHD-3 has three parts. Part one covers the primary headache disorders. Part two covers the secondary headaches, sorted by what is causing them. Part three covers the painful cranial neuropathies and other facial pains, a group that includes trigeminal neuralgia and sits slightly apart from headache proper. Within those parts sit numbered groups, and within the groups sit individual diagnoses, each defined by explicit criteria labelled A through E.

Those criteria are lists of countable things. A given diagnosis might require a minimum number of attacks, a duration window, a set number of pain characteristics out of four listed, and a requirement that no better diagnosis explains the picture. This is deliberately mechanical. It exists so that a clinical trial recruiting people with chronic migraine recruits the same population everywhere, and so that a treatment shown to work in one study can be trusted to apply to the patient in the room.

The language is stiff on purpose, and it stays stiff in every translation. A classification is a working instrument for clinicians and trial designers before it is a reading list for patients, which is why its definitions read like tax code. The logic underneath the criteria is the part a non-specialist can actually use. Primary headaches are sorted by phenotype, meaning the observable pattern of the attacks, because their underlying biology is still only partly understood. Secondary headaches are sorted by cause, because the cause is the point. Once that split is clear, the numbering stops looking arbitrary and starts working as a map: a diagnosis in part one is a description of behaviour, and a diagnosis in part two is a claim about origin.

One more consequence of that design. ICHD-3 includes “probable” versions of most primary diagnoses, for cases meeting all criteria but one. That category exists because real patients routinely fall a little short of a tidy definition, and the classification would rather name that honestly than force a fit.

Primary headaches: when the headache is the whole condition

A primary headache disorder is diagnosed when the headache is not being caused by any separate disease. Part one of ICHD-3 holds four groups: migraine, tension-type headache, the trigeminal autonomic cephalalgias, and a miscellaneous group of other primary headache disorders.

The last of those is a genuinely odd collection, and it is where several headaches that people search for individually actually live. Primary cough headache, primary exercise headache, headache associated with sexual activity, cold-stimulus headache (the ice cream headache, which has a formal entry), external-pressure headache from a tight hat or goggle strap, primary stabbing headache, nummular headache confined to a coin-sized patch of scalp, hypnic headache that wakes older adults at the same hour each night, and new daily persistent headache, which starts on a day the person can name and simply does not stop. Each is defined largely by its trigger or its timing.

The three that account for most of the burden are tension-type headache, migraine, and cluster headache, and they differ from one another far more than their shared label suggests. Tension-type headache is common, bilateral, and comparatively mild. Migraine is a neurological disorder with a sequence of stages, of which pain is one. Cluster headache is a rare disorder of extreme, short, strictly one-sided attacks with visible autonomic signs. Treating them as three intensities of one problem produces bad care, because the drugs that work for one are frequently useless for another.

Two things are worth knowing about all primary headaches. First, no blood test or scan confirms them. The diagnosis is made from history, supported by a normal examination. Second, a normal MRI does not rule anything in. It removes some alternatives from the list, which is a different and more limited achievement than people generally expect from imaging.

What is understood about the biology has moved considerably in the past two decades. The trigeminovascular system, the network of nerve fibers supplying the blood vessels and coverings of the brain, is central to how head pain is generated and relayed. Calcitonin gene-related peptide, a signalling molecule released by those fibers, turned out to matter enough that blocking it became a working migraine treatment. That is about as direct a confirmation of a mechanism as clinical medicine offers. Central sensitization, in which pain-processing pathways become progressively easier to trigger, helps explain why frequent headaches tend to become more frequent.

Secondary headaches: when the pain is a symptom of something else

A secondary headache is one where a separate disorder is producing the pain, and the classification sorts them by that disorder rather than by how they feel. ICHD-3 lists eight groups: headache attributed to head or neck trauma, to cranial or cervical vascular disorder, to non-vascular intracranial disorder, to a substance or its withdrawal, to infection, to disorder of homeostasis, to a disorder of the skull, neck, eyes, ears, nose, sinuses, teeth or mouth, and to psychiatric disorder.

To count as secondary, the headache has to be causally tied to the other condition, not merely coincident with it. The criteria look for evidence: the headache began in close temporal relation to the other disorder, or worsened as that disorder worsened, or improved as it was treated. Someone with lifelong migraine who develops a sinus infection has two things happening, and separating the two is ordinary clinical work, done by asking when each pain started and what happened to it once the infection was treated.

The overwhelming majority of secondary headaches are common and self-limiting. Headache during a flu, a hangover, the pressure headache of a bad head cold, the headache of caffeine withdrawal on the second morning of a holiday, the ache after a night of poor sleep and too little fluid. These are secondary headaches in the formal sense, and nobody investigates them.

The rare end of the category is what drives the caution that surrounds headache assessment. Subarachnoid hemorrhage, arterial dissection in the neck, cerebral venous sinus thrombosis, meningitis and encephalitis, giant cell arteritis, raised intracranial pressure from any cause, and carbon monoxide poisoning all present with head pain. The clinical problem is one of base rates. A great many people have headaches, and a very small number of them have one of these, so the screening question is whether anything in this particular presentation fails to fit the pattern of a familiar headache.

Brain tumor deserves a plain word, because it is the diagnosis most often searched alongside headache. Tumors do cause headache, but headache alone, with a normal neurological examination and no change in pattern, is an uncommon way for one to announce itself. The typical picture involves other findings: a new neurological deficit, seizures, personality change, morning vomiting, or a headache that has changed character and is steadily worsening over weeks.

Tension-type headache: the most common kind

Tension-type headache is the most common headache disorder in the world, and the most reliably under-described. Its defining features are close to the opposite of migraine’s: pain on both sides rather than one, a pressing or tightening quality rather than pulsating, mild to moderate rather than moderate to severe, and, importantly for the diagnosis, not made worse by routine physical activity such as walking or climbing stairs. Nausea is absent. The criteria allow photophobia or phonophobia, sensitivity to light or to sound, but not both together, because that combination points elsewhere.

The subtypes are defined by frequency, and the thresholds matter more than they look. Infrequent episodic tension-type headache occurs on average less than one day a month. Frequent episodic covers one to fourteen days a month. Chronic tension-type headache means fifteen or more days a month, for longer than three months. That last line marks a point at which the disorder itself has changed, and it is where treatment shifts from managing attacks to preventing them.

Many people with tension-type headache have tenderness in the pericranial muscles, the muscles of the scalp, temples, and neck, detectable when a clinician presses on them. The mechanism is not settled. The prevailing account holds that peripheral factors dominate in the episodic form. That means pain arising in the muscles and the connective tissue around them, along with tenderness a clinician can feel by pressing on the same tissues. Central sensitization is thought to take over in the chronic form. The evidence for the muscular contribution is weaker than the disorder’s name implies, and “tension” in the label refers to the quality of the pain rather than to psychological stress or to a proven muscle spasm.

It is the least researched of the common headaches, largely because it is the least dramatic. It rarely sends anyone to an emergency department, it responds acceptably to simple analgesics in its episodic form, and it does not have the visible attack patterns that make migraine and cluster tractable to study. The result is a disorder that affects more people than any other headache type and has the thinnest literature behind it. That gap is worth naming plainly, because it is the reason so much advice about tension-type headache sounds confident and is not.

Migraine: a neurological disorder with phases

Migraine is a disorder of the brain in which head pain is one component of a longer sequence. Global burden estimates put the one-year prevalence at roughly 14 to 15 percent of the world’s population, with national surveys ranging from about 10 to 20 percent depending on how cases are counted. Among adults, two to three women are affected for every man, a ratio that emerges after puberty and narrows again later in life. It runs strongly in families.

The attack has up to four stages. The prodrome, hours to a day or two beforehand, can bring yawning, food cravings, mood change, neck stiffness, and fluid retention, and many people learn to recognize it in themselves before the pain arrives. Aura, present in a minority of attacks, is a wave of fully reversible neurological symptoms, most often visual: a shimmering arc, a zigzag edge, an expanding blind spot. Aura typically develops over more than five minutes and resolves within an hour. It can also be sensory, spreading numbness or tingling up an arm and into the face, or affect speech. The headache phase follows, classically one-sided and pulsating, moderate to severe, made worse by movement, and accompanied by nausea or by sensitivity to light and sound. Untreated, it runs somewhere between four and seventy-two hours. Then the postdrome, the day-after state of depletion and slowed thinking that patients describe consistently and that questionnaires routinely miss.

Aura is thought to correspond to cortical spreading depression, a slow wave of neuronal excitation followed by suppression that moves across the cortex at a rate matching how visual aura expands across the visual field. That correspondence between an animal-model phenomenon and a symptom people report is one of the more satisfying pieces of headache science.

Several named variants sit under migraine in the classification. Migraine with and without aura are the two core diagnoses. Chronic migraine describes headache on fifteen or more days a month for more than three months, with migraine features on at least eight of them. Vestibular migraine, in which vertigo is the dominant symptom and headache may be minor or absent, is a common cause of recurrent dizziness that gets diagnosed late. Hemiplegic migraine involves reversible weakness during aura and has known genetic forms. Migraine with brainstem aura brings symptoms such as double vision, slurred speech, and unsteadiness.

One reclassification is worth stating plainly, because it affects a lot of people who have never seen a neurologist. Most people who arrive with a self-diagnosed sinus headache, meaning facial pressure, nasal congestion, and pain that worsens on bending forward, meet the criteria for migraine when assessed properly. Migraine attacks routinely produce nasal and tearing symptoms through the same autonomic pathways, and there is no infection to find.

Cluster headache and the other trigeminal autonomic cephalalgias

Cluster headache belongs to a group called the trigeminal autonomic cephalalgias, defined by strictly one-sided pain in the trigeminal nerve territory occurring together with autonomic signs on the same side. Those signs are visible to anyone in the room: a red and watering eye, a drooping lid, a constricted pupil, a blocked or running nostril, facial sweating, all on the painful side only.

The attacks are short and extraordinarily severe. Untreated, they last from about fifteen minutes to three hours, and they recur from once every other day up to eight times a day. Where a person in a migraine attack usually wants to lie still in the dark, a person in a cluster attack characteristically cannot keep still, and paces or rocks. That behavioral difference is diagnostically useful precisely because it is so consistent.

The timing is the other signature. Attacks tend to strike at the same hour, often waking the person in the early hours, and in the episodic form they arrive in bouts lasting weeks to months separated by remissions that can run for a year or more. Some people have bouts at the same season each year. That circadian and circannual patterning points toward the hypothalamus, and imaging studies during attacks have supported hypothalamic involvement. In the chronic form, remissions are absent or brief.

Population studies put cluster headache at around 0.1 percent of adults, with individual surveys running from roughly 0.05 to 0.4 percent depending on how cases are found and confirmed. It is more common in men, though the male predominance in recent series is less lopsided than older figures suggested. Diagnostic delay is one of the disorder’s persistent scandals: people commonly spend years being treated for sinus disease or dental problems before the pattern is recognized, and the condition’s other name, suicide headache, reflects how severe the untreated attacks are.

The rest of the group is rarer still and separates mainly on attack length and frequency. Paroxysmal hemicrania produces shorter attacks, minutes rather than an hour, at higher daily frequency. SUNCT and SUNA produce attacks lasting seconds to a few minutes, sometimes dozens or hundreds in a day. Both names are descriptions of the attack. SUNCT stands for short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing, meaning brief one-sided stabs of nerve-type pain with a red, watering eye. SUNA is the same picture with other autonomic signs, such as a blocked nostril, in place of the red eye. Hemicrania continua is a continuous one-sided headache with autonomic features and exacerbations. Two of these, paroxysmal hemicrania and hemicrania continua, are defined in part by responding completely to indomethacin, which makes a supervised trial of that drug both a treatment and a diagnostic test. It is an unusual arrangement in medicine, and it is the reason getting the label right in this group changes everything about what happens next.

Medication-overuse headache and other secondary types worth knowing

Medication-overuse headache, still widely called rebound headache, develops in people who already have a headache disorder and who take acute pain medication frequently over months. The result is headache on fifteen or more days a month, with the treatment sustaining the problem it was taken for. The classification separates drug classes by how many days a month qualify as overuse: simple analgesics such as paracetamol, aspirin, and other anti-inflammatories carry a higher day threshold, while triptans, ergot derivatives, opioids, and combination analgesics carry a lower one. The counting runs over months rather than weeks, and the diagnosis applies only to people who already have an underlying headache disorder such as migraine or tension-type headache. The same pattern of use in someone without one does not produce the syndrome.

The mechanism is incompletely understood, and probably involves changes in central pain modulation with repeated exposure. What is well established is the shape of the problem: a person with episodic migraine treats attacks, attacks become more frequent, treatment becomes more frequent, and the pattern converts to daily headache. Management involves withdrawing the overused drug, with a period of worsening before improvement, alongside preventive treatment. It is one of the few headache diagnoses where the intervention is subtraction.

Several other secondary types come up often enough to be worth naming. Post-traumatic headache follows concussion or other head injury and can persist for months. Cervicogenic headache originates from structures in the upper neck and refers pain into the head, usually one-sided and provoked by neck position. Headache attributed to temporomandibular disorder brings jaw pain and clicking with it. Acute rhinosinusitis genuinely does cause headache, alongside fever and purulent nasal discharge, meaning mucus that has turned thick and yellow or green, but chronic sinus disease without acute exacerbation is a much weaker candidate than its reputation suggests. Uncorrected refractive error and eye strain are blamed constantly and account for a small fraction of what is attributed to them.

Then a shorter list of secondary causes that are uncommon but specifically looked for. Idiopathic intracranial hypertension raises pressure inside the skull without a mass, typically in younger adults, and threatens vision through swelling of the optic discs. Giant cell arteritis inflames the temporal arteries in people over fifty, and can cause irreversible blindness, which is why new persistent headache with scalp tenderness or jaw pain on chewing in that age group is treated urgently. Ordinary high blood pressure is a poor explanation for chronic headache; hypertensive crisis is a different matter. Carbon monoxide poisoning causes headache in everyone exposed in a household, which is a pattern worth knowing about.

How doctors tell one type from another

For primary headaches, the history is the test. A clinician reaches the diagnosis by asking questions in a particular order, and the answers carry more weight than any scan does.

What a clinician is assembling is a pattern in time. The questions usually start with age at first onset, then move to how fast the pain reaches its maximum, which is the single most important thing to establish in an acute presentation. From there the clinician builds a picture of a typical attack: how long it runs untreated, how many days a month it arrives, where the pain sits and what it feels like, and whether ordinary movement such as climbing stairs makes it worse. Next comes the company the pain keeps, meaning nausea, sensitivity to light and sound, a watering eye, a blocked nostril, aura, weakness, or fever. What the person does during an attack is asked for a reason, since lying still in the dark and pacing the room point in different directions. Medication is counted twice over: which drugs are taken, and separately, on how many days a month they are taken. Family history goes in because these disorders run in families. And, running underneath all of it, whether the pattern has changed, because a changed pattern is a different question from a longstanding one.

Headache diaries earn their reputation here. Retrospective recall of headache frequency is unreliable in a consistent direction, and a month of prospective recording routinely produces a different number from the one a patient offered at the first appointment. It also captures medication days, which is how medication-overuse headache is usually caught.

Examination is targeted rather than exhaustive: a neurological examination, blood pressure, fundoscopy to look at the optic discs for signs of raised pressure, palpation of the temporal arteries where age and features raise the question, and assessment of the neck and jaw.

Imaging enters the picture when something in the history or examination raises a specific question. Guidelines across countries agree on the other half of that rule. A typical primary headache pattern with a normal neurological examination does not call for a scan. The reasoning is not only about cost. Scanning unselected populations turns up incidental findings at a meaningful rate. Each of those findings then generates repeat imaging, a specialist referral, and worry, and all three can outlast the headache that prompted the scan in the first place. When imaging is indicated, non-contrast CT is the fast option in the emergency setting, MRI is the more detailed elective study, and vascular imaging of the head and neck is added when dissection or aneurysm is in question. Lumbar puncture follows a negative CT when subarachnoid hemorrhage is still suspected, and inflammatory markers are checked when giant cell arteritis is on the list.

Which symptoms point to a medical emergency?

Emergency medicine works from a published list of headache features that prompt urgent assessment, and it is worth stating once, in its standard form, as the guidance clinicians actually use. The features are:

• A headache that reaches maximum intensity within about a minute, known as thunderclap headache
• Headache with fever, neck stiffness, or rash
• Any new neurological symptom or sign: weakness, numbness, confusion, seizure, visual loss, difficulty speaking, unsteadiness
• A first or new headache after age fifty
• Headache with a history of cancer or a compromised immune system
• Headache that changes with posture, being clearly worse lying down or worse standing up
• Headache brought on by coughing, straining, or exertion
• A headache that is progressively worsening over days to weeks, or a clear change in a longstanding pattern
• Headache following head injury
• Headache in pregnancy or the weeks after delivery
• Swelling of the optic discs found on examination

Thunderclap headache is the one that most deserves elaboration, because the word “sudden” is doing precise work. The defining feature is the time to peak: pain that goes from nothing to its worst within seconds to a minute. The classic description is a blow to the back of the head. The concern is subarachnoid hemorrhage from a ruptured aneurysm, along with dissection, venous thrombosis, and reversible cerebral vasoconstriction syndrome. Many thunderclap presentations turn out to have benign causes, and the investigation happens anyway, because the ones that do not are catastrophic and time-sensitive.

These features are population-level screening criteria. They describe which presentations published guidance routes to same-day or emergency assessment. The list is deliberately over-inclusive: most people who meet one of these features turn out to have an ordinary cause, and the cost of investigating a benign headache is much lower than the cost of missing a bleed.

What doesn’t fit neatly into a type

The categories above are cleaner in the classification than they are in a waiting room, and the honest version of this subject has to say so.

Overlap is ordinary. A great many people have both migraine and tension-type headache, with different attacks fitting different criteria, and the two coexist more often than either occurs alone in headache clinic populations. Medication-overuse headache typically layers on top of one of them rather than replacing it, which is why a person can arrive with three headache diagnoses that all need addressing in sequence. Chronic daily headache, a phrase clinicians still use in conversation, describes a frequency and nothing more, and it usually resolves on assessment into chronic migraine, chronic tension-type headache, medication overuse, or some combination.

Diagnoses also migrate over time. Episodic migraine can convert to chronic migraine and convert back. Episodic cluster headache can become chronic and, less often, the reverse. A headache pattern that has been stable for fifteen years can shift after a head injury, a pregnancy, or a change in medication, and the new pattern gets classified afresh, on its own criteria.

This is also the point at which self-classification runs out of road, and the reason is built into how the criteria are written. Almost every primary headache diagnosis requires counting: a minimum number of prior attacks, a frequency over months, a duration observed without treatment. Those quantities cannot be assessed from inside one headache, which is the moment when people most want an answer. The criteria are built for a pattern viewed from above, over time, with the alternatives systematically excluded, and exclusion is the part that cannot be done from a symptom list.

The evidence base itself is uneven in ways worth naming. Migraine and cluster headache are comparatively well studied, with mechanisms partly mapped and treatments derived from them. Tension-type headache, despite being the most common, has thin evidence behind both its mechanism and its management. New daily persistent headache remains poorly understood, defined mainly by the fact that its sufferers can name the day it started. Several of the classification’s newer entries exist as placeholders for phenomena that are clearly real and not yet explained.

What is settled: the primary and secondary division, the criteria for the major primary disorders, the red-flag features, and the finding that CGRP signalling is central to migraine. What is not: the mechanism of tension-type headache, why headache disorders become chronic in some people and not others, and what determines the timing that makes cluster headache behave like a clock. Questions worth taking to an appointment: does this pattern meet criteria for a specific diagnosis or a probable one, has anything in my history flagged as a reason to investigate further, on how many days a month am I taking acute medication, and what would change the diagnosis if the pattern shifts?