Ask anyone how long a menstrual cycle is and you'll get the same answer a textbook, a sex-ed class, and every period app's default settings give: 28 days. It's on diagrams. It's baked into the pill's 21+7 packet design. It's the number your app assumed about you before you'd logged a thing.

Here's what the number actually is: a rounded-off convenience that most real cycles don't match. When researchers analyzed real-world data from more than 600,000 cycles logged by over 100,000 people (published in npj Digital Medicine, 2019), the results were blunt:

  • Only around 13% of cycles were actually 28 days long
  • The real average was about 29.3 days, but averages mislead here, because…
  • Cycle lengths varied enormously between people (most falling somewhere in the mid-20s to mid-30s), and, crucially,
  • They varied within the same person: month-to-month differences of several days were the norm, not noise. Meaningful ovulation-timing variation persisted even among people whose cycles looked "regular" on paper.

In other words: the population isn't a 28-day clock with a few outliers. It's a wide, wobbly distribution, and you are not a data-entry error for failing to match the diagram.

Where 28 came from, and why it stuck

The 28-day figure is old: a tidy round number (suspiciously lunar, four neat weeks) that early medical literature adopted and everything downstream inherited. It stuck because it's useful to institutions: textbooks need one diagram, pill packets need one schedule, and, most relevantly today, apps need one assumption to show you a due date on day one, before you've given them any data.

None of that makes it true of you. "Average" was never a synonym for "normal," and in cycle data the distinction has teeth: doctors define normal as a range (cycles of roughly 21–35 days, with several days of person-to-person and month-to-month variation), not as proximity to 28 (the full rules on what's normal and what's worth a doctor's visit are here).

The real cost of the myth

This would be harmless trivia if the number stayed in textbooks. It doesn't. It shapes:

  • Worry. A 33-day cycle reads as "5 days late" against the myth (cue the 1 a.m. googling and the unnecessary pregnancy tests) when it's simply a 33-day cycle.
  • App behavior. The myth is load-bearing in most trackers: 28 is the day-one default, and the averaging math that replaces it is just the myth, personalized (why that math fails). The result is software that treats your variability, the most human thing about your cycle, as an error to be smoothed away.
  • Who gets failed hardest. If your cycles genuinely run long or variable (PCOS being the most common reason), the myth graduates from annoying to alienating: every default, diagram, and due date quietly tells you your body is doing it wrong. It isn't. The diagram is.

What deserves to replace it

Not a better number: a better shape. Three replacements, from the data up:

  1. Your range, not the average. "My cycles run 27–34 days" contains more truth than any single number ever will. It's learnable from a few months of logged dates, and it's the version of "regular" that actually describes bodies.
  2. Variation as normal, not noise. A cycle that moves a few days month to month isn't misbehaving: 600,000 cycles say that's what cycles do.
  3. Predictions that carry their own uncertainty. If cycles are a distribution, an honest forecast is a range with a confidence level (wider in noisy months, narrower as data accumulates), not a bold date with an apology tour afterward. That conviction is the entire reason Ebb exists: a tracker built on your range, honest about its confidence, designed for the ~87% of cycles the diagram forgot. (How that compares to the other trackers on the market.)

The 28-day cycle was never a fact about bodies. It was a fact about diagrams. Bodies deserve software, and expectations, built on the data instead.

FAQ

Is a 28-day cycle normal?
Yes, and so is 24, 31, and 35. "Normal" is a range (roughly 21–35 days for most adults), not one number. 28 is simply one ordinary value inside it that history happened to canonize.

What percentage of women actually have a 28-day cycle?
In the largest real-world dataset published, only about 13% of cycles hit exactly 28 days, and even people whose cycles average 28 rarely hit it every month.

Is it normal for cycle length to change every month?
A few days of month-to-month variation is thoroughly normal. Variation beyond about 7–9 days between your shortest and longest cycles is where doctors start using the word "irregular" (common, and worth a conversation if it's new). (Details here.)

Why do apps assume 28 days?
Because they must show something before you've logged anything, and 28 is the culturally installed default. The better question is what an app does after your data arrives: keep averaging, or model your actual range.