How long does it really take to form a habit?
THE SHORT ANSWER
A median of 66 days, not 21. In the only well-known study to measure this directly, participants took between 18 and 254 days for a new daily behaviour to feel automatic, with a median of 66. The 21-day figure comes from a 1960s plastic surgeon’s observation about patients adjusting to their appearance, not from research on habits.
Where the 21-day myth came from
In 1960, a plastic surgeon named Maxwell Maltz published a popular book in which he noted that his patients seemed to take about 21 days to get used to a changed appearance. That observation — about adjusting to a new nose, not about building a habit — was repeated, simplified and eventually detached from its origin entirely. Sixty years later it is still quoted as though it were a measurement.
What happened when someone actually measured it
In 2010, researchers at University College London asked 96 people to pick a new daily behaviour — a glass of water with lunch, a walk before dinner — and to log each day whether they did it and how automatic it felt. Automaticity climbed along a curve that rose steeply at first and then flattened out.
The median time to reach the plateau was 66 days. But the median is the least interesting number in that study. The range ran from 18 days to 254 days, and it depended heavily on the behaviour: a glass of water automated fast, fifty sit-ups did not.
Why the range matters more than the median
If you adopt 66 days as your new rule, you have swapped one arbitrary number for a slightly better arbitrary number. The useful finding is that difficulty predicts duration. A behaviour that takes ten seconds and requires no decision will feel automatic within weeks. A behaviour that requires changing clothes, leaving the house and enduring discomfort may take most of a year — and that is normal, not evidence that you are failing.
The finding almost nobody quotes
Buried in the same study is the most practically useful result: missing a single day had no measurable effect on the formation curve. One skipped day does not set you back and does not reset anything.
This matters because the real danger after a missed day is not the day. It is the reasoning that follows it. Having broken the run, people frequently conclude the attempt is ruined and abandon it altogether — a pattern documented across dieting, exercise and quitting research. The missed day costs you nothing. The story you tell yourself about it costs you the habit.
What to do with this
- Set a horizon of roughly two months, and treat anything before that as too early to judge.
- Pick a behaviour small enough that difficulty is not the variable — you can grow it later.
- Anchor it to something that unmistakably happens each day, rather than to a time.
- When you miss a day, do it again the next day and stop counting. That is the whole recovery protocol.
This is also why any app that wipes your streak after one missed day is working against the evidence. It is optimising for the anxiety that drives engagement, not for your habit.
Sources
- Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W., & Wardle, J. — "How are habits formed: modelling habit formation in the real world", European Journal of Social Psychology (2010)
- Polivy, J., & Herman, C. P. — "If at first you don’t succeed: false hopes of self-change", American Psychologist (2002)
- Maltz, M. — "Psycho-Cybernetics" (1960)
Questions
- Is 21 days enough to form a habit?
- For a very easy behaviour, possibly. For most, no. The measured median is 66 days and the range extends past 250. Twenty-one days is a number from a 1960 book about patients adjusting to plastic surgery, not a research finding about habits.
- Does missing one day break a habit?
- No. In the study that measured habit formation directly, missing a single opportunity had no meaningful effect on how automatic the behaviour eventually became. What breaks habits is abandoning them after a miss.