New habit in 21 days? What the research actually shows

In short

The 21 days are a myth. In the best-known study, it took a median of 66 days for a new behaviour to feel automatic, with a range of 18 to 254 days.
A 2024 review paints a similar picture: about two months is typical, but people differ enormously.
What matters is not the calendar but repetition in the same context. Missing one day won’t set you back.

The problem in everyday terms

You decide to read for ten minutes every evening instead of scrolling. The first week goes well. By week three it still takes effort, and you’ve read somewhere that after 21 days it should run on its own. So you conclude: I’m just not a routine person.

This is exactly where the 21-day rule does harm. It sets a deadline that has little to do with the research. Miss it, and you think you’ve failed, when you are simply in the middle of a normal learning process.

The good news: how long it really takes, and what helps, has by now been studied in some detail.

What the research says

The landmark study: a median of 66 days. A team led by psychologist Phillippa Lally at University College London followed 96 people, mostly students, for 84 days (Lally et al., 2010). Each chose a new eating, drinking or exercise behaviour and tied it to a daily cue, such as a bottle of water with lunch or a 15-minute run before dinner. Every day, they rated how automatic the behaviour felt.

The result was a typical learning curve: automaticity rises quickly at first, then levels off. For the 39 people whose data fitted the model well, it took a median of 66 days to get close to that plateau. The range ran from 18 to 254 days. Exercise took a median of 91 days, drinking 59 and eating 65, but with such small groups this difference was not statistically reliable.

Missing a day barely matters. In the same study, skipping a single day lowered automaticity only slightly and had no lasting effect. The researchers concluded that occasional slips do not meaningfully slow habit formation.

The review: two months, but wide spread. A systematic review and meta-analysis from the University of South Australia pooled 20 studies with 2,601 participants (Singh et al., 2024). Only four of them measured time to habit directly: medians were 59 to 66 days, means 106 to 154 days, and individuals needed anywhere from 4 to 335 days. Factors the review links to stronger habits include frequency, timing, choosing the behaviour yourself, and finding it pleasant. Morning habits and self-chosen habits tended to be stronger.

Big datasets: months or weeks, depending on the behaviour. A study in PNAS used machine learning on more than 12 million gym visits and more than 40 million hospital handwashing opportunities (Buyalskaya et al., 2023). Going to the gym typically took months to become predictable; handwashing in hospitals took only weeks. Again: there is no magic number.

What a habit actually is. In psychology, a habit is an action that a cue in your day triggers automatically, because you have repeated it often enough in exactly that situation (Gardner, Lally & Wardle, 2012). Once that link is strong, the behaviour needs less motivation and often lasts even after the early enthusiasm fades.

What it doesn’t say

There is no fixed number. The 66 days are not a new rule either. They come from a small group, mostly students in London, and only about half of the analysed data fitted the model well. The range of 18 to 254 days is the real message.

Automaticity was self-rated. In almost all of these studies, participants report on a questionnaire how automatic a behaviour feels. That is a well-established tool, but not an objective measure. The 2024 review also rests on relatively few, often small studies; the findings on morning routines or self-choice each come from single studies.

Where the 21 days come from. According to Gardner and colleagues, the myth seems to trace back to anecdotes in Maxwell Maltz’s 1960 book Psycho-Cybernetics. They concerned cosmetic surgery patients who had got used to their new appearance after about three weeks, not people building new habits.

Habit is not the same as discipline. A habit grows from repetition in a specific situation, not from willpower alone. And it grows from doing: not doing something doesn’t become a habit in the same way. If you want to drop a behaviour, replacing it with a new one often works better.

No substitute for professional help. If you’re dealing with addiction, sleep problems or other health issues, a doctor or psychologist will help more than any routine.

How to try it this week

Pick a small, concrete action you can do daily and actually want, such as reading two pages or drinking a glass of water. Small beats ambitious.
Tie it to a fixed cue that happens every day: “After my first coffee, I read two pages.” Set the time and place, in the morning if you can.
Do it every time the cue comes, the same way. While you’re learning, keep the place and form constant.
If you miss a day, just carry on the next. No making up, no self-blame; a single slip changes little.
Expect about ten weeks. Once a week, ask yourself: does it feel easier than at the start? A tick on a sheet of paper is enough to keep you going.

Sources

Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W. & Wardle, J. (2010). How are habits formed: Modelling habit formation in the real world. European Journal of Social Psychology, 40(6), 998–1009. doi.org/10.1002/ejsp.674
Singh, B., Murphy, A., Maher, C. & Smith, A. E. (2024). Time to form a habit: A systematic review and meta-analysis of health behaviour habit formation and its determinants. Healthcare, 12(23), 2488. doi.org/10.3390/healthcare12232488
Buyalskaya, A., Ho, H., Milkman, K. L., Li, X., Duckworth, A. L. & Camerer, C. (2023). What can machine learning teach us about habit formation? Evidence from exercise and hygiene. PNAS, 120(17), e2216115120. doi.org/10.1073/pnas.2216115120
Gardner, B., Lally, P. & Wardle, J. (2012). Making health habitual: the psychology of “habit-formation” and general practice. British Journal of General Practice, 62(605), 664–666. doi.org/10.3399/bjgp12X659466