Menstrual Migraine: Why Estrogen Withdrawal Sets Off So Many Headaches

The headache turns up on a schedule. That is the part nobody notices for years.

It gets filed under stress, because there was a deadline that week. Then under sleep, because there was a bad night. Then under the weather, the glass of wine on Friday, the skipped lunch, the new office lighting. Each of those explanations holds up exactly once. None of them explains why the same thing arrived last month, and the month before, and will arrive again in about four weeks.

Most of my working life has gone into finding bugs that only reproduce under one specific condition. The maddening ones are never actually random. They look random because you are logging the wrong variable. You chase the deploy, the network, the load — and then somebody notices it only happens on the first of the month, and the whole mystery collapses into an obvious answer in about ten seconds.

Migraine has a variable like that. For an enormous number of people, it is hormonal, and it has been sitting in plain sight the entire time.

The pattern hiding in the numbers

Migraine is roughly three times more common in women than in men. That gap is one of the largest sex differences in any common neurological condition, and it does not exist for the first decade of life. Before puberty, boys and girls get migraine at about the same rate. The lines separate at menarche and stay separated for the next forty years.

A timing coincidence that precise is not a coincidence.

Among women who have migraine, somewhere between half and two-thirds report that at least some of their attacks track the menstrual cycle. A smaller group — in the region of one in ten — has attacks that occur almost exclusively in that window and essentially nowhere else in the month.

The International Classification of Headache Disorders gives this a definition tight enough to be useful. Attacks that begin between two days before the start of bleeding and three days after it — a five-day window, written as day −2 to day +3 — in at least two out of three cycles. If they also happen at other times of the month, it is menstrually-related migraine. If they happen only in that window, it is pure menstrual migraine.

Five days. Three cycles. That is a testable claim about your own body, and most people have never tested it.

It is the fall, not the level

The obvious guess is that low estrogen causes migraine. The obvious guess is wrong, and the experiment that showed it is wrong dates back to the early 1970s.

Brian Somerville gave women with menstrual migraine an injection of long-acting estradiol shortly before their periods were due. Menstruation arrived on schedule anyway — the hormone did not delay the bleeding. But the migraine was postponed, and it stayed postponed until the estradiol in their blood finally fell. When he ran the same design with progesterone, the periods were delayed and the migraines were not.

So the trigger is not the altitude. It is the descent. A sustained low level is tolerable; the body adapts to it. What the brain objects to is the drop — the rate of withdrawal after a stretch of high exposure.

Why the drop matters comes down to what estrogen has been doing while it was there. It modulates serotonergic signalling, dampens the excitability of the trigeminovascular system, and interacts with calcitonin gene-related peptide, the neuropeptide that sits at the centre of modern migraine biology. Estrogen is, in effect, holding the threshold up. When it withdraws quickly, the threshold drops with it, and whatever else is going on that week is suddenly enough.

There is a second mechanism stacked on top of the first. Prostaglandins released from the shedding endometrium peak in the first forty-eight hours of bleeding, and prostaglandins are independently capable of provoking headache. This is not an academic footnote — it is the reason anti-inflammatories can work well in this specific window when they do nothing much for the rest of the month.

The clinical consequence of all this is that menstrual attacks behave differently. They tend to last longer. They are rated as more disabling. They respond less reliably to a single dose of acute medication, and they recur within twenty-four hours far more often than non-menstrual attacks. They are usually migraine without aura. If you have ever felt that the period headache is not the same animal as your other headaches, you are not imagining the difference.

The same mechanism, across a whole life

Once you see migraine as a response to hormonal withdrawal rather than to hormone levels, the entire arc of a life reads differently.

Childhood. Rough parity between boys and girls. No monthly cycle, no monthly withdrawal.

From menarche onward. The rates diverge and keep diverging. Each cycle now contains a scheduled estrogen fall.

Pregnancy. Many women improve substantially in the second and third trimesters, when estrogen is high and, crucially, stable — no fall, no monthly trigger. This is genuinely one of the more reliable patterns in the literature, though it is not universal and the first trimester is often rough.

Postpartum. Delivery produces the steepest estrogen drop a human body ever experiences, over a matter of days. Attacks in the first week or two after birth are common, and they arrive at the exact moment a person has the least capacity to absorb them. Almost nobody is warned.

Perimenopause. Usually the hardest stretch of all. The intuition is that migraine should ease as estrogen declines, and the intuition is backwards. Perimenopause is not a smooth downward slope; it is erratic — spikes and crashes, cycles of varying length, unpredictable withdrawal events happening more often than once a month. More falls per unit of time means more triggers per unit of time.

After menopause. For migraine without aura, things frequently improve once the fluctuation genuinely stops. Which means the common reassurance — it will get better after menopause — is often true and still nearly useless, because it skips over the several years of turbulence in between and offers nothing for them.

The one thing worth knowing before anything else

If you take one item from this piece to a doctor, make it this one.

Migraine with aura carries a modestly elevated risk of ischaemic stroke on its own. That baseline risk is small in absolute terms, but it multiplies with combined hormonal contraception — anything containing estrogen — and it multiplies again with smoking. Most national and international guidance, including the World Health Organization's eligibility criteria, treats estrogen-containing contraception as contraindicated in women who have migraine with aura. Progestogen-only methods are generally regarded as acceptable.

Two things make this worth stating plainly. First, plenty of people were prescribed a combined pill years ago and developed aura later, and nobody re-asked the question. Second, aura is often assumed to mean flashing lights, so people who get the other kinds do not report it. Aura can be visual — zigzags, a blind spot, shimmering — but it can equally be tingling that spreads up an arm to the face, or a stretch of not being able to find words, typically building over five to twenty minutes and resolving within an hour.

If any of that is familiar and you are on an estrogen-containing contraceptive, that conversation is worth having sooner rather than at your next routine appointment.

Separately, and regardless of the hormonal picture: a headache that arrives at maximum intensity within seconds, the worst headache of your life, a first migraine after roughly age fifty, fever with a stiff neck, or neurological symptoms that do not resolve are all reasons for urgent assessment rather than tracking.

Why a calendar changes the toolkit

Here is the part that makes tracking worth the effort rather than just satisfying.

If your attacks are unpredictable, there are two treatment strategies available: treat each attack when it arrives, or take a daily preventive every day of the year to reduce how often they arrive. That is the whole menu.

If your attacks are predictable inside a five-day window, a third strategy opens up that does not otherwise exist: short-term prevention. Treat the window, not the calendar year. Take something for about six days a month rather than three hundred and sixty-five, aimed at a threshold you know is about to drop.

That option is only available to people who can demonstrate the pattern. Not assert it — demonstrate it.

There is a second benefit, and in practice it may be the larger one. It changes the conversation you are able to have. "I get bad headaches, quite often, I think maybe around my period" is a sentence that produces a prescription for whatever the clinician usually prescribes. "Here are three cycles. Attacks on day −1 to day +2 in all three. Severity seven to eight. Sumatriptan gets me to functional in ninety minutes but it comes back by hour eighteen every time" is a sentence that produces a different appointment entirely. It is specific enough to be acted on.

Most people are not under-treated because their doctor is careless. They are under-treated because they walked in with an anecdote and walked out with a general answer.

What is genuinely different about treating it

What follows is a map of what exists, so you can ask a real question. It is not a recommendation, and none of it should be started, stopped or adjusted without the person who actually knows your history.

Acute treatment, but earlier and with the recurrence problem named. Because menstrual attacks are more refractory and much more likely to come back within a day, the standard approaches get adjusted — treating at the first sign rather than waiting to see, combining a triptan with an anti-inflammatory, or choosing an agent with a longer half-life so the coverage outlasts the attack. Recurrence is the signature failure mode here, and it is worth reporting as its own symptom rather than as evidence the drug did not work.

Short-term prevention, sometimes called mini-prophylaxis. A short course started a day or two before the expected onset and continued for five or six days. Naproxen and other anti-inflammatories are used here, which makes sense given the prostaglandin surge. Frovatriptan is the most-studied option in this role, largely because its long half-life lets it cover a multi-day window; naratriptan and zolmitriptan have been used similarly. Magnesium started in the second half of the cycle has some supporting evidence. All of it depends on knowing when the window starts, which is why the tracking comes first.

Flattening the fall itself. Rather than treating the consequences of the withdrawal, some approaches remove the withdrawal — continuous or extended-cycle hormonal contraception that eliminates the hormone-free interval, or perimenstrual estradiol delivered by gel or patch to soften the drop. This is powerful when it fits and it is not for everyone; whether it is even on the table depends entirely on the aura question above and on the rest of your cardiovascular picture.

Standard and newer preventives. For people whose menstrual attacks sit on top of frequent attacks the rest of the month, the conversation is about background prevention, including the CGRP-targeting drugs developed over the past several years. Given that estrogen modulates CGRP signalling, there is a satisfying mechanistic logic to that class being relevant here, and the evidence in menstrual migraine specifically is still accumulating.

A tracking template worth bringing to an appointment

Paper, a notes app, or a headache diary app — the medium matters far less than the columns and the three cycles. One row per day, including the days nothing happens. The empty days are the control group; without them you cannot tell clustering from frequency.

• Date
• Cycle day. Day 1 is the first day of full flow. Days before it are negative: −1, −2, −3. This single convention is what makes the pattern visible; a calendar date will not do it.
• Headache today? Yes or no.
• Onset time and roughly when it ended.
• Worst severity, 0 to 10.
• Aura? Type — visual, sensory, speech — and how many minutes it lasted.
• Associated symptoms. Nausea, vomiting, light sensitivity, sound sensitivity, smell sensitivity.
• What you took, the dose, and how long until you were functional. Record it even when it failed. Especially when it failed.
• Did it come back within 24 hours? Yes or no. This column earns its place.
• Hours of function lost. Work, childcare, everything else.
• Confounders. Under six hours of sleep, alcohol, a skipped meal, illness, an unusual stressor.

Three cycles minimum. Two is not enough to meet the definition and one is a story.

What you are looking for at the end: do the attacks fall inside day −2 to day +3, in at least two of the three cycles? If yes, you have a documented menstrual pattern and the short-term prevention conversation is open. If they scatter across the month with no clustering, that is also a real result — it rules something out and points the search elsewhere.

Then compress the three cycles onto a single page before the appointment: attacks per month, how many landed in the window, severity range, what you tried and what it did, how often it recurred within a day, and total hours of function lost. Bring the raw log as backing, but hand over the page. Ten minutes is a short appointment, and a summary someone can read in thirty seconds changes what the other twenty-nine are spent on.

Questions people actually ask

My attacks are around my period but not exactly on it. Does that still count?
Possibly, and it is worth logging rather than guessing. The −2 to +3 definition exists to make research comparable, not to police your experience. Cycles vary, ovulation produces its own smaller estrogen dip that catches some people mid-cycle, and a pattern that sits slightly outside the formal window is still a pattern a clinician can work with. Log it and let the data argue.

I have had migraines for twenty years and no one ever asked about my cycle. Why not?
Partly because a lot of migraine is treated in short appointments where the presenting complaint is the headache and the history stops there. Partly because the person who could most easily spot the pattern is you, and the pattern is only visible over months. It is a genuine gap in routine care rather than a reflection on you, and it is one of the few gaps a patient can close single-handedly with a notebook.

Will menopause fix it?
Often, eventually, for migraine without aura — but the honest version includes the perimenopausal years first, which are frequently worse than what came before. Treating the transition as a stretch that deserves active management, rather than a wait, is the more useful framing.

Is it safe for me to be on the pill?
That depends on whether you have aura, and it is not a question to settle from an article. Combined hormonal contraception is generally contraindicated with migraine with aura because of the stroke-risk interaction; progestogen-only options are usually acceptable. If you have any history of visual, sensory or speech symptoms before an attack, raise it specifically — do not wait to be asked.

Do I have to wait three cycles before saying anything?
No. If attacks are frequent or disabling, go now and start the log in parallel. The three cycles make the conversation sharper; they are not an entry requirement for getting help.

What strikes me about all of this is that the information was never missing. The body was keeping the record the whole time, in a format nobody was reading. Three months of a notebook does not change the biology at all. It just makes the pattern legible to someone who can do something about it.