A Habit in 21 Days? Where the Number Came From and What the Data Say

You've decided to do ten minutes of exercise every morning. The first week is easy, carried by enthusiasm. In the second, you have to talk yourself into it. And somewhere in the third, a nagging thought appears: "This should be automatic by now. Why am I still negotiating with myself every morning?"

That expectation comes with a specific number. "It takes 21 days to form a habit" is a line from motivational posts, habit-tracker apps and "new you in three weeks" challenges. The number sounds convincing: not too long, not too short. The only problem is that research doesn't back it up.

Where the 21 days came from

The trail usually leads to Psycho-Cybernetics, a 1960 book by plastic surgeon Maxwell Maltz. In its preface he wrote that it "usually requires a minimum of about 21 days to effect any perceptible change in a mental image." His examples were patients getting used to a new face after surgery, amputees and people who had moved house. He was talking about adjusting to a new state, not about new actions.

Behaviour researchers at University College London suggest that two different processes got mixed up here. Habituation is when something new stops surprising us. A habit is when an action starts on its own, without a separate decision. Over time, "a minimum of about 21 days" became simply "21 days" and took on a life of its own in popular culture.

What the data say

In psychology, a habit isn't just something we do often. It's an action that happens almost automatically in response to a familiar situation: you walk in the door, you hang your keys on the hook. Researchers measure this automaticity with questionnaires — how much an action happens "by itself," without effort or thought.

The classic study is the work of Phillippa Lally and colleagues, published in 2010. Ninety-six volunteers chose a new eating, drinking or activity behaviour and repeated it daily in the same situation for 12 weeks. Eighty-two of them provided enough data for analysis.

Automaticity rose quickly at first, then levelled off. Among those whose results could be described by the model, the median time to reach that plateau was 66 days. The range ran from 18 to 254 days. The upper end is a model forecast, since the observation itself lasted only 84 days. Simpler actions became habitual faster than more complex ones.

A 2024 systematic review and meta-analysis in the journal Healthcare pooled 20 studies with 2,601 participants. They covered healthy eating, physical activity, flossing, drinking water and other daily behaviours. Only four studies measured how long it actually took to form a habit. Their median values were 59–66 days, the means 106–154 days, and individuals ranged from 4 to 335 days.

An even broader picture came from a 2023 study in PNAS. Researchers from Caltech and other universities used machine learning to analyse two large datasets: gym attendance by more than 30,000 people over four years, and handwashing by more than 3,000 hospital workers during their shifts. Handwashing became a predictable habit within a few weeks. For the gym, the authors first reported about four to seven months, but a later correction revised this to about two to three months (68–78 days). Many news stories still quote the old figure. One of the authors, economist Colin Camerer, said that the 21-day figure "was not based on any science."

So the honest answer to "how long?" is this: usually longer than three weeks, often around two months, but very different for different people and different behaviours.

Why so long — and why so varied

A habit is a link between a situation and an action. Each repetition in the same situation strengthens it a little. That's why a habit is less like a switch that clicks on one day and more like a path through grass: it's worn in gradually, and the moment it's "done" is hard to pin down.

The data hint at what affects the pace:

• frequency. The more often an action is repeated, the faster automaticity grows;
• a stable context. The same situation, the same trigger. In the gym data, the day of the week was one of the strongest predictors of attendance;
• gaps. For 76% of gym-goers, the time since their last visit mattered: the longer the pause, the less likely they were to come back;
• personal choice and enjoyment. In the studies covered by the review, habits that people chose themselves, and found at least somewhat pleasant, tended to stick better;
• complexity. A glass of water is not the same as a run.

On time of day, the data are thin. In one study from the review, morning stretching became a habit faster than evening stretching. Whether that holds for other behaviours is unknown.

Where the explanation falls short

The science of habits is more honest than motivational posts, but it has its own limits.

There isn't much data. Only four studies in the review measured time to habit formation. Many samples were small (the median was about a hundred participants), and the review's authors rated 11 of the 20 studies as having a high risk of bias.

Automaticity is mostly self-reported. People rate for themselves how "automatic" an action has become. That's useful, but imprecise.

Almost all of it is about health. Food, exercise, flossing. We know much less about how habits form around studying, playing an instrument or learning a language.

66 isn't a magic number either. It's easy to turn it into the new "21 days." But it's a median for only part of the participants in one study — those whose data the model described well. The spread matters more than the middle.

What you can take from this

Nobody can promise a habit will arrive by a certain date. But a few calm conclusions follow from the data.

• Don't measure yourself by the calendar. If you still have to talk yourself into it on day 22, that's not failure — it's a normal part of the process.
• One missed day isn't a disaster. In Lally's study, missing a single opportunity did not materially affect habit formation. Long breaks matter more, as the gym data show.
• Attach the action to something already in your day. After your morning coffee, after you lock the door, after lunch. A stable trigger works better than "sometime today."
• Start with a simpler version. Two minutes instead of twenty. A complex action is easier to build once a simple one is already in place.

A small experiment

Pick one very small action and tie it to something that happens every day anyway — say, one chess puzzle with your morning coffee. After a week, don't ask yourself "is it a habit yet?" Ask something else: "Do I still pause to think before I do it?" The disappearance of that pause is the sign that the path is being worn in.

Questions for discussion

1. Which of your habits formed without you noticing, with no plan at all?
2. Have you ever given up on something new because "it should be easy by now"?
3. What works better for you: a fixed time, or tying a new action to another daily one?