How long does it take to form a habit?

Updated 19 August 2026

• 66 days is a median. Half the people in the study took longer than that.
• The spread ran from 18 days to 254. Which end you land on is not knowable in advance.
• The 21-day figure comes from a 1960 book by a plastic surgeon and was never about habits.
• A glass of water settled faster than 50 sit-ups. Size does more work than motivation.
• Missing a single opportunity had no measurable effect on the curve.
• Very little shows in the first week. The fair moment to judge it is somewhere past six.

Where 66 days comes from

In 2010 a team at University College London published the study everyone is quoting when they say a number. Ninety-six people each chose one everyday behaviour — a piece of fruit with lunch, a glass of water, a short run — and committed to doing it daily, in the same situation, for twelve weeks. Once a day they logged whether they had done it, and each week they filled in a short scale measuring how automatic it felt: how much it happened without deciding, without remembering, without effort.

Plotted over time, those scores rise steeply and then level off. The researchers fitted a curve to each person and asked when it reached 95% of its own plateau. The median answer was 66 days.

The median is the least interesting part of the result. The individual curves ranged from 18 days to 254, and 254 was where the modelling stopped rather than where that person did. One participant's curve had not levelled off by the end of the study at all.

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:10.1002/ejsp.674

Where 21 days comes from

The 21-day figure traces to Psycho-Cybernetics, a self-help book published in 1960 by Maxwell Maltz, a plastic surgeon. He noticed that patients seemed to need about three weeks to stop seeing their old face in the mirror, and that people who lost a limb reported a phantom sensation for roughly the same period. His sentence was that it takes “a minimum of about 21 days” for an old mental image to dissolve.

Two things happened to that sentence over sixty years. “A minimum of about” was dropped, and the subject changed from self-image to behaviour. What is left is a number with no study behind it, applied to something it was never measured on.

It survives because it is convenient. Three weeks is a marketable length for a programme, and 66 days is not.

What actually changes the number

How complicated the behaviour is. In the UCL data, drinking a glass of water with lunch became automatic considerably faster than doing 50 sit-ups before breakfast. This is the single most actionable finding in the whole study, and it points the same way as everything else: make it smaller than feels serious.

Whether the situation repeats. Automaticity is built by the same cue producing the same action, over and over. A habit attached to something that happens every single day gets more repetitions per week than one attached to something that happens on weekdays. A habit attached to “when I get a moment” gets almost none, because that cue never actually arrives.

How consistent the surroundings are. Habits are learned associations between a context and a response, so a stable context does most of the teaching. This also explains the common experience of a solid routine falling apart after a house move or a job change: the cues went with the old life.

Not willpower. Nothing in the data supports the idea that people who took 18 days were more disciplined than people who took 200. What separated them was mostly what they had chosen to do and how reliably the moment for it turned up.

What it feels like along the way

Knowing the shape of the curve is worth more than knowing its length, because the shape is what makes people quit at the wrong moment.

Week one. Every repetition is a decision. This is normal and it is not evidence about you or about the habit. Almost nothing has changed yet because almost nothing can have.

Weeks two to four. The steepest part. It starts occasionally happening before you have thought about it, usually on the days when the moment is most ordinary.

Weeks five to ten. The flattening. Progress gets harder to notice from the inside, which is exactly when a progress bar would be discouraging. A better question than “how far along am I” is whether this week cost less than the same week a month ago.

After that. The thing stops being a task. You notice it mostly by its absence, on a day when the moment does not come round.

What to do with this

• Do not judge it at day nine. That is inside the part of the curve where nothing has happened yet, and it is where most attempts are abandoned.
• Pick a size you can do on a bad day. Simpler behaviours settled faster in the data, and a version you can always do is what keeps the repetitions coming.
• Keep the moment fixed and let the size vary. The repetition is what teaches the association. A smaller version at the usual time does more for it than a full version at a random one.
• Do not count days. Nothing in the research says a run of consecutive days matters, and the one thing it does say about misses is that a single one made no measurable difference.

How Mini Loops handles it

Mini Loops does not tell you how far along you are, because the honest answer is a range from 18 to 254 days and a number would be pretending otherwise. What it does instead is describe how the habit is settling in, in words, once a week. It also asks you for a small version at setup, so that on the days when the full thing is not happening there is still something to repeat.

When the same obstacle keeps turning up it says what it noticed and proposes a change to the plan. Nothing changes until you agree to it.