How Long Does It Take to Form a Habit? What the Study Says

This guide explains what that study measured, why its range is so wide, and what it cannot tell you. If you are choosing a tool for the job, our guide to habit tracker widgets is the place to start. If you have just lost a streak, what to do when you break a habit streak is the more practical read. This post is about the evidence.

The short answer, and why it is not 21 days

“21 days” is a figure many people repeat, and it is easy to see why: it is short, tidy and sounds like a promise. The research digest of the British Psychological Society, which summarises the study discussed here, does not mention it. What it reports is an average of 66 days to reach maximum automaticity, and a spread so large that the average is a poor guide for any one person.

“Automaticity” is the key word. In this research a habit is not “something you did for N days”. It is the degree to which a behaviour has become something you do with little thought. The study tracked that feeling day by day, which is why the answer is a curve and not a deadline.

What the study measured

The paper is Lally, van Jaarsveld, Potts and Wardle (2010), published in the European Journal of Social Psychology. Here is the design in plain words, as reported in the paper’s abstract and the research digest.

• Who: 96 volunteers. The digest describes them as undergraduates with a mean age of 27.
• What they did: each person chose one eating, drinking or activity behaviour and did it every day, in the same situation, for 12 weeks (84 days).
• How it was measured: every day they recorded whether they had done the behaviour, and rated how automatic it felt with a questionnaire called the Self-Report Habit Index.
• The analysis: for each person, the researchers tried to fit a curve through the daily scores. Automaticity rose quickly at first and then levelled off. The time to reach 95% of that plateau (the “asymptote” in the paper) was the measure.
• The data: 82 of the 96 volunteers supplied enough data for analysis. The curve could be fitted for 62 of them, and 39 of those fits were good.

The result is the one in the opening lines: a range from 18 to a predicted 254 days, and an average of 66 days reported by the digest.

Why the range is so wide

The 254-day end deserves a careful reading. The study lasted 84 days, and many people were still improving when it stopped, according to the digest. The digest explains that the longest figure is a projection that assumes the still-rising rate of change continues beyond day 84, not something anyone was observed to reach. The digest adds that about half of the participants had not reached a high enough automaticity score by day 84.

Two things push the range wide:

1. The behaviour. The digest reports that exercise behaviours took about one and a half times as long to reach their plateau as eating or drinking behaviours. A reasonable reading, though not a finding of the study, is that a behaviour needing more effort has more to overcome.
2. The person. Volunteers chose their own behaviour and their own situation. The abstract also reports that performing the behaviour more consistently was associated with a better fit of the curve, so consistency may be part of the picture.

So the 66 days is the middle of a crowd, not a target. Treat it as “usually weeks to months”, which is a more honest and more useful statement.

What a missed day did

This is the finding most worth remembering. According to the paper’s abstract, missing one opportunity to perform the behaviour did not materially affect the habit formation process. The digest puts it as a single missed day having little impact on later automaticity gains, and notes that William James argued in 1890 that a behaviour must be repeated without omission to become a habit, which this result suggests overstated the cost of a slip.

It also gives the other half. Many missed repetitions, even spread out, had a cumulative effect: they lowered the level of automaticity people reached. The lesson is about patterns. One gap is not a reset; a long run of gaps is what costs you.

What the study can and cannot tell you

Two cautions follow. First, this is one study with a modest group, and the 66-day figure is an average over a very uneven spread. Second, the behaviours were everyday eating, drinking and activity choices. The study says nothing about quitting something, and nothing about health outcomes. This post makes no health claims and gives no medical advice. If you want help with a health matter, ask a qualified professional.

A practical reading: small, daily, with room for gaps

You can use the findings without turning them into rules the study did not set.

• Make the behaviour small. The digest reports that exercise behaviours took longer than eating or drinking ones. Choosing a smaller version you can do on a bad day is a fair design choice, though the study did not test it.
• Pin it to a situation. Participants did the behaviour in the same context each day. “After my coffee” is a clearer cue than “sometime today”.
• Expect weeks to months. If day 30 does not feel automatic yet, that is a normal place to be, given the range above.
• Plan for gaps. One missed day made little difference; many mattered more. So the useful question after a slip is “when is my next repetition?”, not “have I failed?”.

None of this proves that any one routine will work for you. It describes what happened on average in a group.

How a tracker can help without nagging

A tracker does not form the habit. It records the repetitions, and it can make a single slip less dramatic. That second part is where design matters, because a counter that goes to zero on one missed day teaches the opposite of what the study found.

Spareday is a streak and habit tracker that treats a slip differently. Each habit gets three spare days a month, which keep a streak going across a day you missed, and you choose when to use one. You can also skip a day on purpose, or fill in any of the last 366 days if you forgot to tap. The free version covers up to five habits, widgets, reminders and statistics, and it needs no account. You can read the details on the Spareday page in the Store.

To be clear about the limits: the study did not test spare days, skip days or any app. Spareday’s design is a choice that is consistent with one finding, namely that a lone slip is not the end, and it is not a result of that research. If you want to know what you keep and what you give up when a tracker needs no account, see habit tracker with no account: what you keep and lose.

Questions

How long does it really take to form a habit?

There is no fixed number. In a widely quoted study the average was 66 days for a daily behaviour to reach its peak automaticity, and individual results ran from 18 days to a predicted 254 days. Expect weeks to months, longer for behaviours that take more effort.

Is the 21-day rule true?

The research digest does not mention it. In the study, the average was 66 days and the range very wide, so the study gives no support for 21 days as a deadline. Some people got there sooner, many took longer, and about half of the volunteers had not reached a high enough score after 84 days.

What is the 66-day figure?

It is the average time for 96 volunteers to reach their highest level of self-reported automaticity for a daily eating, drinking or activity behaviour, as reported in the research digest. It describes a group in one study, not a promise for you.

Does missing a day reset the habit?

No. In the study, one missed day had little effect on later automaticity gains. Many missed repetitions, even spread over a long time, did lower the level people reached, so the pattern of gaps is what matters.

Do some habits take longer?

Yes. The digest reports that exercise behaviours took about one and a half times as long as eating or drinking behaviours to reach their plateau. The paper also found that doing the behaviour more consistently gave a better fit of the automaticity curve.