Key facts
- 96 volunteers repeated a daily behaviour of their choice for 12Â weeks and rated each day how automatic it felt. 1
- Automaticity levelled off after 66Â days on average; individual times ranged from 18 to about 254Â days. 3
- Simple actions such as drinking water became automatic faster than elaborate routines such as 50 sit-ups. 3
- A single missed opportunity did not seriously disrupt the process. 3
- The 21-day idea traces to the 1960 preface of Psycho-Cybernetics by plastic surgeon Maxwell Maltz. 2
What counts as a habit in research?
In their 2012 review in the British Journal of General Practice, Benjamin Gardner, Phillippa Lally and Jane Wardle describe a habit as an action triggered automatically by contextual cues that were linked to the action through past repetition. Once the link exists, the action depends less on conscious attention and motivation, so a habit can last after the urge fades.
What did the study by Lally and colleagues find?
The study, published in the European Journal of Social Psychology (online 2009, October 2010 issue), followed 96 volunteers for 12Â weeks. Each chose an eating, drinking or activity behaviour to repeat daily in the same context and rated every day how automatic it felt. The researchers then fitted a curve to each person's ratings.
| Item | What is reported | Source |
|---|---|---|
| Participants | 96 volunteers; 82 with usable data | Crossref; BPS Research Digest |
| Follow-up | 12Â weeks (84Â days) | Crossref; BPS Research Digest |
| Measure | Daily self-rating of how automatic the behaviour felt | BPS Research Digest |
| Time to level off | 66Â days on average | UCL blog; Gardner, Lally and Wardle (2012) |
| Range | From 18Â days to about 254Â days (an estimate for people who had not levelled off by day 84) | Crossref; BPS Research Digest |
| Model fit | The individual curve fitted well for 39 of the 62 people modelled | Crossref |
How to read the 66-day figure
An average describes a group; it is not a deadline for any one person. In the study, the typical time to reach a plateau was 66Â days, but individual times were spread over a very wide range: the quickest participant levelled off after 18Â days, while the slowest estimate was about 254Â days. That last value is an extrapolation for people whose curves had not flattened by the end of the 84Â days of follow-up, so it is a projection rather than something anyone was observed doing.1
Two details help put the number in context. First, the volunteers chose their own behaviour and repeated it daily in the same context, so the figure describes people doing exactly what the model assumes. Second, according to the British Psychological Society, about half of the volunteers had not reached a habit-level automaticity score after 84Â days, which is one reason the slowest times had to be estimated.4
Where does the 21-day rule come from?
A University College London blog post by Benjamin Gardner and Susanne Meisel (29 June 2012) traces the figure to the 1960 preface of Psycho-Cybernetics by Maxwell Maltz, a plastic surgeon turned psychologist. Maltz described adjustments that take about three weeks, such as getting used to a new face or a new home. The post's authors call the figure unfounded as a habit-formation time: getting used to a situation is not the same as forming a habit. In their journal review, the same researchers advise clinicians to tell patients to expect about 10Â weeks rather than 21Â days.
What makes a habit easier or slower to form?
In the 2012 review, simple actions such as drinking a glass of water became automatic faster than elaborate routines such as 50 sit-ups. The authors advise choosing a new, small behaviour that is meaningful to the person and tied to a stable daily cue, such as an existing routine. Varying the behaviour or the context works against automaticity, and a habit cannot be formed by abstaining from something.
- A small, specific behaviour rather than a vague goal.
- A stable cue: same time, same place, or right after something already established.
- Repetition in the same context, which is what builds the link between cue and behaviour.
Turning the advice into a plan
The 2012 review by Gardner, Lally and Wardle turns the findings into advice for clinicians, and the same points work for anyone choosing a habit to build.3 The table below is an illustration of how to apply them, not a result from a study.
| Question | Example of an answer | Where the idea comes from |
|---|---|---|
| What is the action? | Drink a glass of water | Simple actions became automatic faster than elaborate ones3 |
| Which cue starts it? | Right after I pour my coffee | A stable daily cue, such as an existing routine3 |
| Where does it happen? | In the same place each day | Repetition in the same context builds the link3 |
| How long should I expect? | Plan on about ten weeks, and expect differences between people | The authors suggest ten weeks rather than 21Â days23 |
| What if I miss a day? | Resume at the next opportunity | A single missed opportunity did not seriously disrupt the process1 |
Notice what the plan leaves out: it sets no goal for how the action should feel, and it does not require a perfect streak. The studies read for this page measure how automatic an action becomes, so a plan built on them is about repetition and cues, not about willpower on a given morning.
Does missing a day ruin a habit?
Not according to the study data: an occasional missed opportunity did not seriously disrupt the process, and automaticity kept rising at the next repetition (Gardner, Lally and Wardle, 2012; Crossref). The researchers also note that more consistent performance went with a model that matched the data more closely (Crossref).
What this study cannot tell you
- It gives no duration that holds for everyone: the authors call it unwise to attach a single number to the process (UCL blog).
- Automaticity was self-reported by the participants, which limits precision (BPS Research Digest).
- About half of the participants had not reached a habit-level automaticity score after 84Â days (BPS Research Digest).
- The model fitted well for only 39 of the 62 individual curves modelled (Crossref).
- It covers simple daily behaviours chosen by the participants. It says nothing about what a behaviour changes for health, or about what should become a habit.
Where to go next
The same research underlies the other guides in this topic. For the vocabulary, read ritual, routine, habit: what is the difference. To see how a cue works at the start of the day, read the guide to building a morning routine. For a daily moment that rewards attention, see how to read a food label. All of them are collected in habits and routines, and you can browse every topic in the Chutex guides.
When to ask for advice
To change a behaviour that matters to you, a doctor, pharmacist or dietitian can help you choose a realistic goal. This page summarises research on habits and does not replace medical advice.
Frequently asked questions
Does it really take 66Â days to form a habit?
Not necessarily. Sixty-six days is the average observed in the study by Lally and colleagues at University College London: depending on the person and the behaviour chosen, automaticity levelled off after anywhere from 18 to about 254Â days. The authors consider it unwise to give a single figure that holds for everyone.
Is the 21-day rule true?
No. According to a 2012 blog post by University College London behavioural scientists, the figure traces to the preface of Psycho-Cybernetics, a 1960 book by surgeon Maxwell Maltz about adjusting to change. The researchers call it unfounded as the time it takes to form a habit, and suggest expecting about ten weeks.
Do simple behaviours become automatic faster?
Yes, according to the 2012 review by Gardner, Lally and Wardle in the British Journal of General Practice: drinking a glass of water became automatic faster than an elaborate routine such as 50 sit-ups. The authors therefore advise starting with a small, specific behaviour that is meaningful to you.
What happens if I miss a day?
In the study by Lally and colleagues, an occasional missed day did not seriously disrupt the process: automaticity resumed rising at the next repetition. The researchers do note that more consistent practice went with clearer automaticity curves among the volunteers followed over the 12Â weeks of the study.
How do I choose a cue for a habit?
The 2012 review in the British Journal of General Practice advises tying the new behaviour to a stable cue in your day, for example right after a routine you already have, and then repeating it in the same context. That repetition is what gradually builds the link between the cue and the behaviour.
Does this study apply to everyone?
No. It covers 96 volunteers who chose a simple behaviour themselves and rated how automatic it felt every day for 12Â weeks. It gives a useful order of magnitude, but it cannot predict how long a given behaviour will take for a given person, so treat the figures as a rough guide.
Sources
- Lally P, van Jaarsveld CHM, Potts HWW, Wardle J. How are habits formed: Modelling habit formation in the real world. Eur J Soc Psychol. 2010;40(6):998-1009 (online 2009). Metadata and abstract read through Crossref; full text not read, Wiley (via Crossref). Retrieved 25 October 2026
- Gardner B, Meisel S. Busting the 21 days habit formation myth. Health Chatter, University College London, 29 June 2012, University College London. Retrieved 25 October 2026
- Gardner B, Lally P, Wardle J. Making health habitual: the psychology of 'habit-formation' and general practice. Br J Gen Pract. 2012;62(605):664-666, PubMed Central. Retrieved 25 October 2026
- British Psychological Society, Research Digest: article on how habits form, British Psychological Society. Retrieved 25 October 2026