Ask someone whether their cycles are irregular and the answer usually describes a feeling rather than a measurement. They are all over the place. They are not what they used to be. Some months nothing happens.
There is a definition, it is in the 2023 international guideline, and it is made of numbers. Knowing them changes two things: whether the phrase applies to you at all, and what you are able to say in an appointment instead of “all over the place”.
The numbers
The definition is not one rule. It changes with how long it has been since your first period, because the years after menarche are not expected to look like the years after that.
| Time since first period | What counts as irregular |
|---|---|
| First year | Nothing — irregular cycles are normal here, as part of the pubertal transition |
| 1 to 3 years | Shorter than 21 days or longer than 45 days |
| More than 3 years, to perimenopause | Shorter than 21 days or longer than 35 days, or fewer than 8 cycles a year |
| More than 1 year, at any age | Any single cycle longer than 90 days |
| No periods at all | None by age 15, or more than 3 years after breast development began |
Two things in that table are worth pausing on.
The first year is explicitly exempt. Not “usually fine” — the guideline treats irregularity in that year as part of the transition rather than as a finding. A great deal of worry happens in that window.
The window narrows as you go. Between one and three years, 45 days is inside the definition of regular. After three years, the same cycle is outside it. Nothing about the body changes on an anniversary; the guideline is describing a settling process and drawing the line where most of it has finished.
The eight-cycles rule is the one to remember
Of all the numbers above, fewer than eight cycles in a year is the one that can be answered without arithmetic.
Cycle length requires two dates and a subtraction, done repeatedly, and it is where logging usually breaks down — particularly when a cycle is long enough that the start of it has been forgotten by the time the next one arrives. Counting how many times it happened in a year requires only a list of dates.
It is also the version of the question a clinician can act on immediately. “Six periods last year” is a finding. “They are all over the place” is a description that needs ten minutes of reconstruction before it becomes one.
A long cycle and a missing cycle are different things
The 90-day rule exists because of how those months actually feel from inside. A cycle that runs very long does not announce itself; it is a series of weeks in which nothing happens, each one individually unremarkable, and the threshold is crossed without a moment that marks it.
The guideline puts a number on it so there is a point at which the waiting stops being something to interpret. More than a year past a first period, a single cycle over 90 days meets the definition of irregular on its own — not as a pattern, not as a trend, as one event.
This is also the rule most often missed by people whose cycles are otherwise fine. One very long cycle in a year of ordinary ones still meets the definition, and is still worth a conversation.
Irregular is a finding, not a diagnosis
Meeting any of these definitions means one thing: ovulatory dysfunction, which is one of the three Rotterdam criteria. A PCOS diagnosis needs two of three, and requires conditions that mimic PCOS to have been excluded first.
Thyroid disease, raised prolactin and several other things produce irregular cycles. That is precisely why the finding opens an assessment rather than closing one, and why a blood panel usually follows a conversation about dates.
The reverse also holds, and gets less attention. Regular cycles do not rule PCOS out. They mean one criterion is unmet; the other two can still be. Plenty of people are diagnosed on hyperandrogenism and ovarian morphology with a cycle that arrives like clockwork.
What most articles get wrong
They give one number. Usually 21 to 35 days, with no mention that the window is wider between one and three years past a first period, and that the first year has no rule at all. Applied to a seventeen-year-old, the adult number turns a normal transition into a problem.
They treat irregular as a synonym for PCOS. It is one criterion of three, and the guideline asks for exclusions before any of them count.
They skip the eight-cycles version. Which is a shame, because it is the only form of the question that someone can answer from memory with any accuracy, and the only one that survives the long cycles where day-counting falls apart.
What this changes about what you write down
If the definitions are made of dates, then dates are the thing worth keeping — and specifically the day bleeding started, which is the only event in a cycle with an unambiguous beginning.
Every figure in the table above is derived from a list of those dates. Cycle length is the gap between two of them. The eight-cycles rule is a count of them. The 90-day rule is one gap. None of it requires recording cycle lengths, which is useful, because a length written down is a number you can no longer check, while a date you can always recalculate from.
That is the whole of the practical advice here, and there is more on how to keep it when the months are unpredictable enough that the usual apps start guessing.