How Long Does It Take to Break a Habit? The 21-Day Number Came From a Surgeon, Not a Study

Short answer: there is no fixed number, and the one everybody quotes is not from research. The only study to time habit formation day by day in ordinary life found a median of 66 days, with individuals ranging from 18 to 254 days. A 2023 analysis of 12 million gym visits and 40 million hospital hand-washing events found that gym habits took months to form while hand-washing took weeks, which is to say the answer depends on the behaviour. Breaking a habit is not the reverse of that clock. It is a different process, driven less by time and willpower than by the cues around you, and the most reliable way to break one that has ever been measured is a change of context.

The 21 days comes from Maxwell Maltz, a plastic surgeon, who observed in his 1960 book Psycho-Cybernetics that patients seemed to take about three weeks to get used to a new face. It was an anecdote about adjustment. It has been repeated as a law of behaviour change ever since.

What was actually measured

Lally and colleagues (2010) at University College London asked 96 volunteers to pick one new daily behaviour, tied to a daily cue, and to report every day for 12 weeks how automatic it felt. Eating a piece of fruit with lunch, drinking a glass of water after breakfast, running for 15 minutes before dinner. Automaticity rose along a curve that climbed quickly at first and then flattened, and the researchers defined "habit formed" as the point where it reached 95% of its plateau.

Three things from that table survive most retellings badly. The 66 is a median, not a rule: half the sample took longer. The range is more than tenfold, which means "how long" has no useful single answer. And a single lapse did not reset anything, which has been confirmed repeatedly since and is the most useful finding in the study for anyone trying to change.

Buyalskaya and colleagues (2023) took the question to a different scale, applying machine learning to gym check-in records covering more than 12 million visits and hand-washing compliance records from hospital workers covering more than 40 million events. The two behaviours formed habits on different timescales: hand-washing in weeks, gym attendance in months. Their conclusion was that the search for a magic number is misguided because the number is a property of the behaviour, not of people.

The time it took participants to reach 95% of their asymptote of automaticity ranged from 18 to 254 days.

Lally, van Jaarsveld, Potts & Wardle, European Journal of Social Psychology, 2010

Why breaking is not forming in reverse

The habit literature draws a distinction that the self-help literature mostly does not. Forming a habit means building a link between a cue and a response until the cue triggers the response without a decision. Breaking one means stopping a link that already fires automatically, and you cannot un-learn an association by wanting to. The old link stays. What changes is whether the cue is present, and whether something else fires instead.

Wood and Neal's (2007) model makes this concrete. A habit is a cue-response association stored in memory, triggered by context, and largely independent of your current goals. That last part is why willpower is the wrong tool: goals and intentions operate on the decision, and a habit runs before the decision happens. You do not decide to reach for your phone when you sit down. The sitting down does it.

Which gives three levers, in descending order of how much they have been shown to matter:

Remove or change the cue. No trigger, no response. This is the mechanism behind the most effective intervention in the literature, below.

Make the response harder or a different one easier. Friction works on habits because habits are effort-minimising by nature. The screen-time article on this site is built around this principle for one specific habit.

Monitor yourself in the moment. Quinn and colleagues (2010) had people report on their real attempts to stop bad habits and found that vigilant monitoring, actively watching for the cue and catching the response as it started, was the strategy most associated with success. Distraction and avoiding the temptation were less effective. This is not willpower in the usual sense; it is attention aimed at the trigger.

The strongest lever is a change of circumstances

The single most reproducible finding in habit research is that habits break when the context that cues them disappears. Psychologists call it the habit discontinuity effect.

Wood, Tam and Witt (2005) followed students transferring to a new university. Their habits of exercising, reading the newspaper and watching television carried over only when the cues at the new place matched the old one: same kind of location, same people around. When the context changed, the habit weakened and behaviour fell back under the control of intentions, which is exactly the window in which a person can choose differently.

Verplanken and Roy (2016) ran a field experiment on sustainable behaviours with households that had recently moved home versus households that had not. An intervention with information and small incentives changed behaviour among the recent movers and did nothing measurable for everyone else. Moving had unlocked the habits; the same message landed on people whose routines were open and bounced off people whose routines were closed.

The practical version, for people who are not about to move house, is to change the cues you can change:

None of these require resisting anything. They remove the trigger or put a competing response in its path.

The habit loop, and where reward fits

The popular model is cue, routine, reward. The research supports the first two strongly and the third more loosely. Rewards matter enormously in forming a habit, which is why sugar, nicotine, notifications and caffeine produce habits so reliably. Once a habit is automatic, though, it will keep running whether or not the reward is still there or still wanted. That is the defining feature of a habit, and it is why "I don't even enjoy it any more" is such a common description of one.

The implication for breaking a habit is that removing the reward is weaker than removing the cue. People keep doing rewarded-then-unrewarded habits for a long time. What they stop doing is habits whose trigger has gone.

Two examples with real timelines

Caffeine. Habitual caffeine is partly a habit and partly a physiological dependence, and the two clocks are different. The withdrawal syndrome runs two to nine days. The habit, reaching for the cup at the usual cue, persists for weeks after the withdrawal has ended, which is why people who have successfully quit still order coffee out of reflex a month later. Keeping the cue and swapping the contents, a decaf in the same mug at the same time, is the habit-research approach, and it is the reason tapering advice usually includes it.

Exercise. In both Lally's sample and the gym data, exercise was the slow one, and the gym study put it at months rather than weeks. If your goal is to break the habit of not exercising, the honest expectation is a season, not three weeks, and the strongest predictor in the gym data of eventually becoming a regular was not motivation but whether the person went on the same days at the same times.

What predicts relapse

The habit-formation studies give a few reliable signals about what pulls people back.

Returning to the old context. Holiday habits die at home and home habits die on holiday, in both directions. If you broke a habit while circumstances were different and then circumstances return, expect the cue to fire again, and plan for it rather than being surprised.

Stress and fatigue. Under load, behaviour shifts toward whatever is most automatic. This is also true of good habits, which is the best argument for building them: they survive the bad week. Bad ones do too.

All-or-nothing thinking. Lally's data showed a single missed day had no measurable effect on the curve. The people who did poorly were the ones who missed many days, and in practice the mechanism is often that one miss is interpreted as failure and becomes a week. The correct response to a lapse is the next scheduled repetition, not a restart.

Relying on the decision. If the plan is "I will just not do it", the plan is being executed by the part of you that a habit bypasses. Plans that work specify the cue and the replacement.

What to actually do

Name the cue, not the habit. Write down where you are, what time it is, who is there and what you just did when the habit fires. That is the thing you are changing.

Change what you can about that context. Location of objects, order of routine, route through the house. The intervention is physical before it is mental.

Put a replacement in the cue's path. Same trigger, different response. A habit vacuum gets filled by the old habit.

Watch for the trigger. Vigilant monitoring, in the research, beat distraction and avoidance.

Use transitions. A move, a new job, a holiday, even a new phone: these open a window of a few weeks in which habits are unusually changeable. It is the best time to make changes and the worst time to expect old routines to protect you.

Expect months for anything with effort in it. Weeks for the simple things, and a lapse is a data point, not a verdict.

Questions people ask

How long does it take to break a habit? There is no fixed number. The research on habit formation found a median of 66 days and a range of 18 to 254, and behaviours that take effort, like exercise, form over months rather than weeks. Breaking a habit depends more on removing the cue that triggers it than on elapsed time.

Is it true that it takes 21 days to break a habit? No. The figure comes from a 1960 self-help book by a plastic surgeon describing how long patients took to adjust to a changed appearance. No study has found 21 days to be a meaningful threshold.

Does missing a day reset the habit? No. In the study that tracked daily automaticity, a single missed day had no detectable effect on the curve. Missing many days in a row did. The problem with a lapse is the decision that follows it, not the lapse itself.

Why is it so hard to break a habit with willpower? Because a habit is a cue-response link that fires before a decision is made. Willpower operates on decisions. It is the wrong tool for a process that runs upstream of it, which is why changing the cue works better.

What is the fastest way to break a bad habit? Remove the trigger, or change your surroundings so the trigger is absent. In field studies, habits weakened most reliably when people moved to a new context, and interventions worked on recent movers that failed on everyone else.

Does it take longer to break a habit than to form one? They are different processes rather than the same one in reverse. The old association is not erased; it is bypassed. A habit can be dormant for years and fire again the moment the original cue returns.

How long does it take to form a new habit? A median of 66 days in the one study that timed it, with wide variation. Simple actions with a clear daily cue form fastest; exercise was slowest in both the small study and the large gym dataset.

Does the habit loop of cue, routine, reward hold up? Cue and routine are well supported. Reward matters for forming a habit but much less for maintaining one, because an established habit runs whether or not the reward is still present.

Why do old habits come back on holiday or after a move? Because the cues that ran them return, or, on holiday, because the cues that ran your good habits are absent. The context is doing the work in both directions.

This article summarises behavioural research for general information. Habits involving alcohol, nicotine, other substances or compulsive behaviours can involve physical dependence, and support from a clinician makes a measurable difference to success.