Day 21 Came and Went and Your New Habit Still Feels Like Work. Science Says That’s Normal

It is the fourth week. The trainers are by the door where you left them, the alarm went off at 6.30, and you are lying there negotiating with yourself all over again.

You were promised this would be automatic by now. Twenty-one days, everyone said.

Here is the reassuring part: nothing has gone wrong. The 21-day figure was never measured in anyone. When researchers finally did measure it, the typical answer was closer to two months, and for some behaviours, going to the gym among them, it stretches to half a year.

In the past three years, two big pieces of research have sharpened that picture: a 2023 machine-learning study that tracked more than 30,000 gym members, and a 2024 review that pooled 20 studies and 2,601 people. Together they explain not just how long a habit takes, but why your handwashing became automatic in a fortnight while your running plan still needs a pep talk.

A surgeon’s hunch became a slogan

The 21 days come from Maxwell Maltz, an American plastic surgeon who wrote the 1960 self-help bestseller Psycho-Cybernetics. He noticed that his patients seemed to need a minimum of about 21 days to get used to a new face, and that people took roughly as long to adjust to other changes in their lives.

Notice the word minimum. Over decades of retelling, it fell away, and an observation about adjusting to change turned into a guarantee about building routines.

Maltz was not running an experiment. He was describing what he saw in a clinic. That is a fine place to start a question and a poor place to end one.

What 96 volunteers in London actually showed

The first serious attempt to time a habit came from Phillippa Lally and colleagues at University College London, published in the European Journal of Social Psychology in 2010. They recruited 96 volunteers and asked each to pick one new eating, drinking or activity behaviour to do once a day in the same situation: a bottle of water with lunch, a piece of fruit, a 15-minute run before dinner.

For 12 weeks, participants logged in every day and rated how automatic the behaviour felt, using statements such as whether it was something they did without thinking. The team then fitted a curve to each person’s scores and looked for the point where it levelled off.

The median time to reach that plateau was 66 days. Individual estimates ran from 18 days to 254.

Two details matter more than the headline number. First, the early repetitions did the most work: automaticity rose fastest at the start and then crept up. Second, the longest estimates sit well beyond the 84 days the study ran, so they are extrapolations from the curve, not observed endpoints. It was also a small sample of mostly motivated volunteers, which is why the newer, bigger studies below are so useful.

Two months is a median, not a finish line

A median of 66 days means half of the people in Lally’s study got there faster and half took longer. It is a midpoint, not a deadline you have personally missed.

The effort mattered too. A glass of water tended to become automatic sooner than 50 sit-ups before breakfast, which makes intuitive sense: the more steps, sweat and willpower a behaviour asks for, the longer the brain takes to hand it over to autopilot.

Your habit is not late. It is just a different size.

The 2024 review that pooled 2,601 people

For years, almost every article on this topic leaned on that single London study. In December 2024, Ben Singh, Andrew Murphy, Carol Maher and Ashleigh Smith at the University of South Australia published the first systematic review and meta-analysis of the question in the journal Healthcare.

They gathered 20 intervention studies, published between 2008 and 2023, covering 2,601 participants aged from their early twenties to their seventies. The behaviours included exercise, flossing, drinking water and taking vitamins. Across the studies that reported timing, median habit formation ranged from 59 to 66 days, while the means were far higher, at 106 to 154 days. Individual times ran from 4 days to 335.

That gap between the median and the mean is the interesting bit. It tells you a minority of people take a very long time, dragging the average up, while the typical person sits near two months.

The review also pulled out what seemed to help. Simple behaviours such as flossing or drinking water strengthened faster than complex ones such as diet and physical activity. Habits people chose for themselves grew stronger than ones they were assigned. Enjoying the behaviour mattered for exercise. Planning a specific time and place helped, as did a stable setting, and so did small “preparatory” routines such as laying out exercise clothes the night before. In one included study, stretching practised in the morning became more automatic than stretching in the evening.

When the review reached newspapers in January 2025, Singh told CNN that the point was to give people realistic expectations: “It’s not often a quick fix.” He also flagged the main weakness of the evidence, that much of the research was short term, so we know less about whether these habits last for years.

What 12 million gym visits revealed

Questionnaires ask people how automatic something feels. A team led by Anastasia Buyalskaya, now at HEC Paris, with Colin Camerer at Caltech and Katherine Milkman and Angela Duckworth at the University of Pennsylvania, took a different route: they watched what people actually did.

Their study, published in PNAS in April 2023, used badge-swipe records from the US chain 24 Hour Fitness covering more than 30,000 members and around 12 million gym visits over several years. A second dataset came from hospitals where more than 3,000 staff wore RFID badges that logged every handwashing opportunity.

Instead of asking whether a habit had formed, the researchers used machine learning to scan hundreds of possible cues, from the day of the week to the time since the last visit, and asked a sharper question: at what point does this person’s behaviour become predictable from context? Predictability, in their framing, is the fingerprint of a habit.

The answer for the gym was months. Caltech’s summary of the paper puts the average at about six months. For handwashing on hospital shifts, it was weeks; HEC Paris reported that most staff got there within around nine to 10 shifts.

The cues were revealing. For 76% of gym-goers, time since their last visit predicted whether they would turn up: the longer the gap, the less likely they were to return. For 69%, the day of the week mattered, with Mondays and Tuesdays the strongest. Time of day, which many of us obsess over, made little difference. At the hospitals, the best predictor was whether someone had washed their hands properly on their previous shift.

It is worth being clear about what these two kinds of study measure. Lally and the 2024 review relied on people rating how automatic a behaviour felt. The PNAS team never asked anyone anything; it inferred habit from how predictable the swipes became. That the two approaches broadly agree, weeks for simple acts and months for demanding ones, is more convincing than either on its own. Neither, though, proves that a predictable routine will survive a new job or a baby.

Why the gym takes months and the sink takes weeks

Handwashing is short, frequent and fused to a fixed cue: you leave a patient’s room, the dispenser is right there. It happens dozens of times a shift, so the repetitions pile up fast, and the action itself is a simple chain of movements.

A gym visit is the opposite. It happens a few times a week at most. It needs packing a bag, travelling, changing and sustained effort, and it competes with work, weather and tiredness. The authors suggested that even well-established gym habits keep needing some planning and intention, in a way that handwashing does not.

That fits the 2024 review’s finding that simple behaviours form faster than complex ones. It also explains why people often feel like failures at exercise specifically. They are measuring a months-long project against a three-week myth.

There was a twist with practical consequences, too. When the gym members took part in a separate programme designed to boost attendance with small incentives, the nudges worked better on the less predictable gym-goers than on people whose routines were already set. Once a habit is established, it seems to run somewhat independently of rewards. That is exactly what you want in the long run, and exactly why the early weeks need more scaffolding.

Miss one day. Just not two

Lally’s team found that missing a single opportunity did not materially derail the habit-forming curve. The automaticity built up so far did not vanish overnight.

The 2023 gym data adds a caveat. Because the gap since the last visit was such a strong predictor for most members, one missed session is harmless, but a growing gap is how routines quietly die. A useful rule of thumb follows from both studies: a lapse is a blip, and the second lapse is the one to guard against.

The bigger danger is psychological. People who see one slip as proof the whole effort has failed tend to abandon it. The data says the opposite: the progress is still there if you turn up next time.

Moving house is a secret reset button

Habits are less about willpower than about where you are. In a 2002 diary study in the Journal of Personality and Social Psychology, Wendy Wood, Jeffrey Quinn and Deborah Kashy asked students to log what they were doing through the day. Around 35% of entries in one study and 43% in the other were habits, things done almost daily and usually in the same place, often while thinking about something else.

A follow-up by Wood, Leona Tam and Melissa Guerrero Witt, published in 2005, tracked 115 students transferring to a new university. Habits such as exercising, watching TV and reading the paper survived when the setting stayed similar, but were disrupted when it changed. Once the old cue was gone, behaviour shifted towards what students actually intended to do.

That is the hopeful flip side. A new job, a house move or even the start of a month loosens the grip of old routines, and the fresh-start effect suggests we are more motivated at those moments anyway. If you are going to begin something, begin it while the furniture of your life is being rearranged.

A 66-day plan that expects wobbles

Put the research together and a realistic plan looks less like a sprint and more like a season. Here is one way to structure it.

Days 1–14: make it tiny and tie it down

Pick one behaviour you actually want to do, since self-chosen habits grew stronger in the 2024 review. Shrink it until it feels almost silly: two minutes of stretching, one glass of water, shoes on and out the door. Attach it to something that already happens every day, written as a rule: “After I put the kettle on, I do ten squats.” This kind of if-then planning gives the brain a cue to latch on to.

Days 15–40: protect the context

Keep the time, place and trigger as stable as you can. Use preparatory habits: put your kit by the bed, the way you might plan a week of outfits from a capsule wardrobe so the morning involves no decisions. If it is exercise, fix the days of the week rather than fretting about the hour; the gym data suggests weekday rhythm matters more than clock time.

Days 41–66: grow it slowly and watch the gaps

Once the small version feels automatic, add a little. Keep an eye on the gap since you last did it rather than on your streak. If you miss a day, do the smallest version the next day, without making up for it. Investing in a few comfortable, hard-wearing basics you enjoy putting on can help too: enjoyment was one of the clearer predictors of exercise habits in the review.

After day 66: expect it to keep changing

For a glass of water, two months is generous. For a running or gym routine, the 2023 data suggests you may still be building at six months, so judge progress by how much less you have to argue with yourself, not by whether it feels effortless yet.

The quiet sign it is working

Habit researchers measure automaticity, but you can feel it without a questionnaire. It is the evening you reach for your trainers before you have decided to go. It is the morning you realise you already did the thing and cannot remember starting.

It rarely arrives on day 21. It may not arrive on day 66. But it does arrive, usually sooner for small things and later for big ones, and almost always for the people who kept the gaps short.

The calendar was never the point. The repetitions were.

Sources

• Buyalskaya, A., Ho, H., Milkman, K. L., Li, X., Duckworth, A. L., & Camerer, C. (2023). What can machine learning teach us about habit formation? Evidence from exercise and hygiene. Proceedings of the National Academy of Sciences, 120(17), e2216115120. DOI
• California Institute of Technology. (2023, April 17). No magic number for time it takes to form habits [News release]. Link
• Korber Hoffman, J. (2025, January 27). Forget 21 days. Most healthy new habits take at least two months to stick. CNN (via KION). Link
• 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
• Maltz, M. (1960). Psycho-Cybernetics. Prentice-Hall.
• Singh, B., Murphy, A., Maher, C., & Smith, A. E. (2024). Time to form a habit: A systematic review and meta-analysis of health behaviour habit formation and its determinants. Healthcare, 12(23), 2488. DOI
• Wood, W., Quinn, J. M., & Kashy, D. A. (2002). Habits in everyday life: Thought, emotion, and action. Journal of Personality and Social Psychology, 83(6), 1281–1297. DOI
• Wood, W., Tam, L., & Guerrero Witt, M. (2005). Changing circumstances, disrupting habits. Journal of Personality and Social Psychology, 88(6), 918–933. DOI

Updated 18 October 2026: we expanded this article with newer research, fresh figures and new sections. Cover photo: Ngo Ngoc Khai Huyen on Unsplash. Charts and graphics: Something Navy.

How we reported this. We went back to the original research for this story, 8 sources in all, listed above, and checked every figure against them. If you think we have got something wrong, tell us and we will look at it. Read more about how we work.