If you have ever quit a new routine on day 22, disappointed that it did not feel automatic yet, you were sabotaged by a number that was never true. The "21 days to a habit" rule is one of the most repeated pieces of self-improvement advice on the internet, and it is wrong. Not slightly off. Wrong at the root, from a source that had nothing to do with habits.
Here is the real picture, drawn from the actual research, and what it means for how you should build a habit if you want it to last.
The short answer
On average, it takes about 66 days for a new behaviour to become automatic. That figure comes from the most-cited real-world study on habit formation, run by Phillippa Lally and colleagues at University College London and published in the European Journal of Social Psychology in 2010. But the average hides the important part. In that same study, the time to automaticity ranged from 18 days to 254 days depending on the person and the behaviour. So a more honest answer is: somewhere between three weeks and eight months, and probably longer than you hoped.
If you take one thing from this article, take this: you are not behind schedule. There is no schedule. There is only a range, and you are somewhere in it.
Where the 21-day myth actually came from
The 21-day figure did not come from a study of habits at all. It came from a 1960 book called Psycho-Cybernetics by Maxwell Maltz, a plastic surgeon. Maltz noticed that his patients seemed to take about 21 days to get used to a change in their appearance, an amputee adjusting to a phantom limb, a patient adjusting to a new nose. He wrote, roughly, that it "requires a minimum of about 21 days for an old mental image to dissolve and a new one to jell."
Read that carefully. He said a minimum, he said about, and he was describing psychological adjustment to a physical change, not the formation of a new daily behaviour. Over the following decades the caveats fell away. "A minimum of about 21 days" became "21 days," and "adjusting to a new self-image" became "building any habit." A surgeon's clinical observation got laundered into a law of behaviour change that millions of people now quote as fact.
This matters because the myth is not harmless. A false deadline sets you up to interpret a completely normal timeline as failure.
What the real study actually did
The Lally study is worth understanding in detail, because the detail is where the useful findings live.
The researchers recruited 96 people who each chose one new health behaviour they wanted to turn into a habit. Some picked eating a piece of fruit with lunch. Some picked drinking a glass of water. Some picked going for a run before dinner. Each person committed to trying to do their chosen behaviour every day, in the same context, for 12 weeks.
Every day, participants logged whether they had done the behaviour and completed a short questionnaire measuring how automatic it felt. Automaticity is the real target here. A habit is not just a thing you do. It is a thing you do without deciding to, triggered by a context rather than by willpower. The researchers tracked how that sense of automaticity grew over time and fit a curve to it for each person.
Three findings came out of that data, and all three should change how you approach habits.
Finding 1: The average was 66 days, but the range was enormous
The median time for a behaviour to reach its automaticity plateau was 66 days. Not 21. And the spread was huge, from 18 days at the fast end to 254 days at the slow end, with one behaviour projected to take even longer than the study could measure.
The takeaway is not "aim for 66 days." It is that the number depends heavily on what you are building and who you are. Which leads to the second finding.
Finding 2: Harder habits take much longer
The complexity of the behaviour mattered a lot. Drinking a glass of water after breakfast automated quickly. Doing 50 sit-ups before dinner took far longer. This is obvious once stated, but people routinely ignore it. They give a daily meditation practice and a daily 5k the same mental timeline, then feel like a failure when the run is still hard at week six while the meditation clicked at week three.
If your new habit is demanding, front-load your patience. It is going to sit at the long end of the range, and that is normal, not a sign it will never take.
Finding 3: Missing one day did not derail the process
This is the single most liberating result in the habit literature, and almost no habit app respects it. The researchers found that missing one opportunity to perform the behaviour did not materially affect the habit formation process. A slip was, in their words, absorbed. The automaticity curve barely flinched.
What did hurt was missing many days in a row, which is a different thing entirely. One miss is noise. A pattern of misses is a trend. The curve cares about the trend, not the single day.
Why this quietly breaks most habit apps
Now hold that third finding next to how habit apps are actually designed. The dominant feature in the category is the streak: a counter that rewards you for never missing and resets to zero the instant you do.
The streak is built on a premise the research contradicts. It treats one missed day as catastrophic, when the science says one missed day is nothing. Worse, it converts a normal event into an emotional failure. You did your habit 25 out of 30 days, which is a genuinely strong month, and the app shows you a red zero and a broken chain. The number erases the truth and hands you a reason to quit.
People rarely abandon a habit on the day they miss it. They abandon it the day after, when the tool tells them the miss meant something. The streak did not protect the habit. It killed it.
A tool that took the Lally findings seriously would do the opposite. It would treat a slip as absorbed, keep showing you the honest signal (days done out of the last 30, a gently decaying score) and make the next day feel doable instead of pointless. This is a deliberate design choice in KeepMeWise, and it comes straight out of this research: the slip is normal, the recovery is the whole game.
So how long will your habit take? A realistic framework
You cannot know your exact number in advance, but you can place yourself in the range with reasonable confidence.
- Simple, once-daily, pleasant, anchored to an existing routine. Think a glass of water after your morning coffee. Expect the fast end, roughly three to five weeks.
- Moderate effort or a new context. A ten-minute tidy, a short journaling practice. Expect the middle, roughly two to three months.
- High effort, physically or mentally demanding, or requiring a big schedule change. Daily exercise, a serious study routine. Expect the long end, three to eight months, and plan accordingly.
The mistake is not being slow. The mistake is quitting during the normal slow part because you were promised fast.
A protocol that respects the evidence
Here is how to build a habit in a way that lines up with what the research actually found, rather than the myth.
- Pick one habit at a time. Automaticity is built through repetition of a specific behaviour in a specific context. Splitting your attention across five new habits slows all of them.
- Anchor it to an existing cue. The Lally participants who tied their behaviour to a consistent context automated more reliably. Use an if-then anchor: "after I brush my teeth, I meditate for two minutes." This is the implementation-intentions effect (Gollwitzer, 1999), and it is one of the best-supported tools in behaviour change.
- Shrink it until it is almost too easy. Early repetitions are where the automaticity curve climbs. A habit you can do on your worst day keeps the curve rising. You can grow it later.
- Track completion, not a streak. Watch days done out of the last 30, or a decaying automaticity-style score. A metric that survives a miss keeps you honest without punishing you.
- Follow one rule after a slip: never miss twice. One miss is absorbed. Two in a row is the start of the trend that actually hurts. Protect the day after a slip above all others.
- Expect 66-plus days, and stop checking for a finish line. The habit is not "done" on a date. It fades in as automaticity rises. If it still takes some effort at week eight on a hard habit, you are exactly on track.
What about the other numbers you have heard?
You will also see "18 days," "30 days," "90 days," and "254 days" quoted as if each is the real answer. They are all just points on the Lally range being cited in isolation. 18 was the fastest single participant. 254 was among the slowest. Neither is the number, because there is no single number. Treat any confident, specific "it takes exactly N days" claim as a signal that the speaker has not read the study.
The bottom line
Building a habit takes about two months on average, often longer for anything hard, and a single missed day does not set you back. The most damaging idea in popular habit advice is not the wrong average. It is the belief that one slip ruins everything, because that belief, not the slip, is what makes people quit. Choose a method and a tool that treat the slip as normal, keep the honest picture in front of you, and let the timeline be as long as it actually is.
Frequently asked questions
Is it 21 or 66 days to form a habit? Around 66 days on average, according to Lally et al. (2010). The 21-day figure traces back to a 1960 book about patients adjusting to changed appearances, not to any study of habit formation.
What is the longest a habit can take to form? In the study, up to 254 days, and one behaviour was on track to take even longer. Harder and less frequent behaviours sit at the long end of the range.
Does missing one day ruin a habit? No. The research found a single missed day was absorbed with almost no effect on the habit forming. What hurts is missing several days in a row, so the practical rule is never miss twice.
Why do habit apps use streaks if streaks are flawed? Streaks are simple and satisfying while unbroken, which makes them popular. The problem is that they treat a normal one-day slip as a failure and reset to zero, which pushes people to quit. A rolling completion metric reflects the evidence better.
How can I build a habit faster? You cannot skip the biology of automaticity, but you can move toward the fast end by keeping the habit small, anchoring it to a fixed daily cue, and doing it in the same context every time. Complexity is the main thing that lengthens the timeline.
Keep reading
- Why streaks backfire, and what works better (why one missed day should not reset you)
- How to build a habit that survives a missed day (the never-miss-twice rule)
- How to set implementation intentions that stick (the if-then plan that beats willpower)