How Long It Takes to Form a Habit
Not 21 days, and not really 66 either. The published research puts it between 4 and 335 days, and the most-quoted figure rests on 39 people. What the evidence actually supports.
Key findings
Each one links to the full figure and the source it came from.
- 4 to 335 days how long habit formation took across the published research
- 59 to 66 days the median times reported, where a median was reported at all
- 39 people the sample the famous 66-day figure rests on
- 18 to 254 days the range inside the study that produced the 66-day figure
- “No.” the study’s own author, asked whether it takes 66 days
- 1960 the year the 21-day rule was published, in a book about plastic surgery
The two numbers everyone quotes
Two numbers dominate this question. One says 21 days. The other says 66 days. Both are quoted as rules, and neither is one.
The 21-day figure is not a research finding at all. It comes from the opening pages of a self-help book, and the man who wrote it was describing something other than habits.
The 66-day figure is real research, and it is a median. The study behind it reported people taking anywhere from 18 to 254 days, and the researcher who produced the figure has said, plainly and recently, that it should not be used as a rule.
What the evidence supports is a range and a list of things that shift the odds. That is less satisfying than a number, and it is more useful, because the number people are given sets an expectation that most of them will miss.
Where 21 days came from
The 21-day rule traces to one book: Psycho-Cybernetics, published in 1960 by Maxwell Maltz, a plastic surgeon. Maltz was not studying habits. He was writing about self-image, and about how long it took his patients to stop being startled by their own reflection.
It usually requires a minimum of about 21 days to effect any perceptible change in a mental image. Following plastic surgery it takes about 21 days for the average patient to get used to his new face.
Maxwell Maltz, Psycho-Cybernetics, 1960
That is the whole basis for it. One clinician's observation about getting used to a new face. There was no experiment, no comparison group, and no measurement of anyone's habits. In the retelling the figure also lost its hedge: Maltz wrote a minimum of about 21 days, and what travelled was "21 days".
What the 66-day study actually found
The 66-day figure comes from a study by Phillippa Lally and colleagues, published in 2010. Volunteers picked an eating, drinking or activity behaviour, did it daily in the same context, and rated how automatic it felt each day for 84 days. The researchers fitted a curve to each person's ratings and read off the point where automaticity levelled out.
The sample shrank at every step. 96 people enrolled. 82 supplied enough daily data to use. The model fitted 62 of those, and 39 of them fitted well. The famous median describes that last group: 39 people, about 40.6% of everyone who signed up.
The range inside that group is the part that rarely gets quoted. The fastest participant reached the plateau in 18 days. The slowest took 254, longer than the study itself ran, so that figure is an extrapolation from the curve rather than an observation. 236 days separate the two.
The clearest answer available comes from the person who produced the number. Asked in 2026 whether it really takes 66 days to form a habit, Lally said:
"No. I mean for someone somewhere for one habit yes, but for most people most of the time, no."
Phillippa Lally, who led the 66-day study
The range across all the evidence
The best current summary of the field is a systematic review published in 2024. It pulled together 20 studies covering 2,601 people. The times those studies reported ran from 4 days to 335 days, a spread of 331 days.
Most of the 20 studies gave no central figure at all. Among the few that did, the medians were 59 to 66 days and the means were 106 to 154 days.
The gap between the medians and the means is itself the finding. A median is the middle person. A mean is pulled upward by anyone at the slow end, and here it is pulled a long way, which means a minority of people took very much longer than the middle person did. Any single headline number has to pick one of those two groups to describe and ignore the other.
The review also pooled 12 studies and 19 intervention conditions into one effect size for habit-forming interventions: 0.69, with a confidence interval from 0.49 to 0.88. That is worth knowing, and it answers a different question. It says that deliberately setting out to build a habit does raise how automatic the behaviour feels. It says nothing about how long that takes, because an effect size is not measured in days.
What makes a habit form faster
The same review looked at what separated the faster habits from the slower ones. The factors it found are the practical part of this report.
Morning practice formed stronger habits than evening practice for the same behaviour. Habits people chose for themselves formed more strongly than habits assigned to them. Enjoying the behaviour helped. So did deciding the time and place in advance, preparing for it in advance, and doing it in the same context each time.
The behaviour itself mattered too. Drinking-water habits formed more strongly than exercise habits. That is a hint about difficulty rather than about willpower: an easy behaviour gets more repetitions, and repetition is the mechanism. If you are picking a first habit, pick a small one.
| Factor | What the review found |
|---|---|
| Time of day | Habits practised in the morning formed more strongly than the same habits practised in the evening. |
| Who chose it | Habits people chose for themselves formed more strongly than habits assigned to them. |
| Type of behaviour | Drinking-water habits formed more strongly than exercise or fruit and vegetable habits. |
| Enjoyment | People who found the behaviour enjoyable formed stronger habits. |
| Planning | Planning the specific time and place in advance was associated with stronger habits. |
| Preparation | Preparatory routines, such as laying out exercise clothes, were associated with stronger habits. |
| Consistent context | Doing the behaviour in the same context each time was associated with stronger habits. |
These are associations across studies of volunteers, not controlled tests of one factor at a time. They show what tends to go with a habit forming faster. They do not prove that moving something to the morning will make it stick.
Every published time to habit formation
Every study in the review that reported a central time to habit formation, with the range each one found.
| Study | Measure | Time to habit | Range reported |
|---|---|---|---|
| Lally and colleagues (2010) | Median | 66 days | 18 to 254 days |
| Keller and colleagues | Median | 59 days | 4 to 335 days |
| Fournier and colleagues, morning stretching | Mean | 106 days | Not reported |
| Fournier and colleagues, evening stretching | Mean | 154 days | Not reported |
behind the numbers
How this was made, and what it can’t do
How each figure was worked out, what it can and cannot support, how to cite it, and every source we used.
Why we publish this
Meadow is a habit app, so we have an obvious interest in this question and an obvious temptation to answer it with a clean number. There is not one to give. The published range is 4 to 335 days, the most-quoted figure describes 39 people, and the researcher behind it says it is not a rule. We would rather say that than repeat something that sets people up to feel behind on day 21.
What the evidence does support is smaller and more usable. Choose the habit yourself. Keep it small enough to repeat on a bad day. Do it at a consistent time, earlier in the day if that is available to you. And do not need it to be fast, because for a lot of people it is not. That is how Meadow is built. Finding the time to repeat anything at all is the harder problem for most people, which is what we looked at in The Self-Care Time Gap.
Cite this report
Free to use with credit. Please link back to this page so readers can check the sources and see any later corrections.
according to Meadow's How Long It Takes to Form a Habit report (https://www.growyourmeadow.com/data/how-long-to-form-a-habit)
Meadow. (2026). How Long It Takes to Form a Habit. Wildcard Wellness. https://www.growyourmeadow.com/data/how-long-to-form-a-habit
"How Long It Takes to Form a Habit." Meadow, 10 Sept 2026, https://www.growyourmeadow.com/data/how-long-to-form-a-habit.
<a href="https://www.growyourmeadow.com/data/how-long-to-form-a-habit">How Long It Takes to Form a Habit</a> by Meadow
The tables and figures compiled here are released under a CC BY 4.0 licence. The underlying survey data is published by the U.S. Bureau of Labor Statistics and is in the public domain.
How we did this
- This report is built on peer-reviewed research rather than survey data, so its sources are published papers and its figures are transcribed from them.
- The 2024 systematic review and meta-analysis by Singh and colleagues is the anchor. It searched the literature systematically, included 20 studies covering 2,601 participants, and reports both the times individual studies found and a pooled effect size for habit-forming interventions.
- The 2010 study by Lally and colleagues is reported separately because it is the source of the widely quoted 66-day figure, and because its own sample sizes are the point.
- Where a figure reaches us through the 2024 review rather than from the original paper, the table says so. We have not re-analysed anything.
- The 21-day claim is traced to its original publication rather than to any of the many pages that repeat it.
Where each figure comes from
- 59 to 66 days - the median times reported, where a median was reported at all
- Medians and means answer different questions here. The means are far higher than the medians because a minority of people take very much longer, which is exactly the detail a single headline figure hides.
- 39 people - the sample the famous 66-day figure rests on
- Lally and colleagues fitted an asymptotic curve to each person’s daily automaticity scores over 84 days. The median of 66 days describes the participants whose curve could be modelled, not everyone who took part.
- 0.69 - the pooled effect size for habit-forming interventions
- A standardised mean difference of 0.69 means the average person after the intervention scored about 0.69 standard deviations higher on habit strength than before it. It says interventions work; it says nothing about how long they take.
Limits of this data
- There is no single answer to how long a habit takes, and this report does not give one. The published range is 4 to 335 days, and the honest summary is that it depends on the person, the behaviour and the context.
- The 66-day median describes the participants in one 2010 study whose automaticity curve could be modelled. It is not a population estimate, and it was never intended as a rule.
- Habit formation studies are mostly small, mostly short, and mostly recruit volunteers who are already motivated enough to sign up. That makes them better at showing what is possible than at showing what is typical.
- "Habit" in this research means automaticity, usually measured with a self-report scale. That is a narrower thing than the everyday sense of the word, and it is self-reported.
- The determinants in the second table are associations found across studies. They point at what tends to help. None of them is a controlled test of that factor on its own.
- The studies cover health behaviours such as eating, drinking, exercise and flossing. Whether the same timings hold for other kinds of habit is not established here.
Update log
Next update: When a newer systematic review of habit formation is published.
- First published, on Lally and colleagues (2010), the Singh and colleagues systematic review (2024), and a 2026 interview with the lead author of the 2010 study.
