How long does perimenopause last? What the averages hide

The most-quoted answer is about four years. The largest study to track the question with menstrual calendars puts the median anywhere from roughly 4.4 to 8.6 years, depending on how old a woman was when it began — and the honest version of the answer adds that the finish line can only be recognized in hindsight, never as it arrives.

Where “about four years” came from

The number has a real source. The Massachusetts Women’s Health Study followed 2,570 women aged 45 to 55 for five years and produced what were, at the time, the first stable estimates of the transition: a median age of about 47.5 at the start of perimenopause, and a transition lasting roughly three and a half years.

Nothing about that finding was wrong. What happened next is the ordinary fate of a good median — it got copied forward, rounded up, and reprinted until it read like a schedule. The “two to ten years” line you meet everywhere is downstream of work like this, flattened.

The number that actually answers the question you are asking

A later analysis took a harder route. Researchers followed 1,145 women across four sites of a long-running multiethnic cohort who kept menstrual calendars — the record that lets you date each stage as it happens rather than asking women to remember it years later.

Then they sorted the women by how old each was when her transition began. The adjusted median duration ran from about 4.4 years in the oldest-onset quarter to about 8.6 years in the youngest-onset quarter.

Sit with the size of that gap. It is close to double, inside a single study, from the same measurements. Earlier onset was associated with a longer transition — so a woman who noticed her cycles changing at 43 and a woman who noticed at 51 were never on the same clock, and the median that sits between them describes neither of them well.

This is the site’s recurring point, in its sharpest form. A population median is a summary of many different curves, and you are living on exactly one of them.

”It” is at least three different clocks

Part of why the answers you find conflict is that “it” is doing a lot of work in the question.

There is the cycle-defined transition. The staging framework clinicians and researchers use marks the early stage by a persistent difference of seven days or more between consecutive cycle lengths, and the late stage by a stretch of sixty days or more without a period. That clock ends at the final menstrual period.

There is the symptom clock, which does not agree with it. In the same large cohort, frequent hot flashes and night sweats lasted a median of 7.4 years in total, and persisted a median of 4.5 years past the final period. The split by timing was even starker: women whose symptoms started while they were still premenopausal or early in the transition had a median duration of 11.8 years or more, while women whose symptoms began after their final period had a median of 3.4 years.

And there is the definition clock, which we will come to, and which is the reason the other two can never be read forward.

Most articles quietly answer the first clock while their readers are asking about the second. If what you want to know is how long will I feel like this, the cycle-based four-year figure was never the relevant number.

Why the endpoint only exists in hindsight

Menopause is dated to a final menstrual period — but a period only becomes the final one after twelve months have passed with no other. Nobody, including you, can identify it on the day. The late stage of the transition, the one that opens with that sixty-day gap, begins on average around two years before the date that will eventually be labeled the last one.

So the forward-looking question — how far along am I? — has no measurement behind it. Not a blood test, not a wearable, not a specialist’s best guess. It is a property of how the endpoint is defined, not a gap in your record or a failure of anyone’s attention. Being told “we can’t say” is, in this one case, the accurate answer rather than a brush-off.

What a watch can and cannot add to that timeline

Nothing on your wrist stages the transition. The criteria are bleeding dates, and they need a calendar rather than a sensor — which is why tracking perimenopause with an Apple Watch starts with the dated log and treats the readings as a second layer.

What the watch adds is a different kind of timeline: continuous, dated, and about you rather than about a cohort. Overnight readings are estimates, not lab measurements, and each one is compared against your own recent nights — the wrist temperature signal page sets out where that relative reading holds and where it does not. Once cycles turn irregular, the ovulation-related shift often stops appearing at all, which cycle tracking in perimenopause works through in detail.

That leaves you with a comparison that is smaller than the one you wanted and considerably more usable: this stretch of weeks against your own previous stretch. Not am I halfway through, but are my nights running warmer than they did in spring — a question your own record can actually answer, and one night sweats and wrist temperature covers signal by signal.

How Perigee reads it

Perigee does not estimate where you are in the transition and will not offer a countdown, because the underlying research does not support one. It compares each signal with your own recent baseline, keeps your logged dates and symptoms beside the readings, and looks for what repeats across weeks rather than what looked convincing on a single night.

When a stretch has no readable pattern, it says so instead of manufacturing one. The research behind these signals sets out what the studies do and do not support, including the parts where the honest answer is that nobody knows yet.

One small thing

Write down the month your cycles first started behaving differently — an approximate month is fine — and keep it somewhere you will find again. It is the single input most strongly associated with how long this lasts, it is the detail memory blurs first, and it is the one thing a clinician can use that no sensor will ever recover for you.