How Long Does It Take to Form a Habit? What the 66-Day Study Actually Found

Habit

The most quoted answer, 21 days, has no study behind it. The second most quoted answer, 66 days, does, but it rarely comes with the details that make it meaningful. In the study behind that number, the time for a new daily behavior to become automatic ranged from 18 to 254 days, about half of the participants never got there within the 12 weeks they were tracked, and exercise took roughly one and a half times longer than eating or drinking habits. The honest answer is a range, and the factors that move you along it matter more than any single number.

Where 21 Days Came From

The 21-day figure traces back to a 1960 self-help book, not to research on habits. In the preface to Psycho-Cybernetics, the plastic surgeon Maxwell Maltz wrote that it usually takes a minimum of about 21 days for a change in a person’s mental image to register, noting that patients needed about three weeks to get used to a new face after surgery. Ben Gardner, a health psychologist then at University College London, traced the myth to this passage in 2012 and pointed out the two details lost in retelling: Maltz was describing adjustment of self-image, not habit formation, and he called 21 days a minimum.

Colin Camerer of Caltech, a co-author of one of the largest habit studies to date, put it bluntly in 2023: the 21-day estimate “was not based on any science.”

The Study Behind 66 Days

The 66-day figure comes from Phillippa Lally and colleagues at University College London, published in the European Journal of Social Psychology in 2010. It was one of the first studies to follow habit formation as it happened in daily life rather than measuring habits people already had.

Element What the study did
Participants 96 volunteers recruited at a university (mean age 27)
Task Choose one eating, drinking or activity behavior and do it once a day in the same context, such as eating fruit with lunch or running for 15 minutes before dinner
Duration 84 days (12 weeks)
Measure Daily self-rated automaticity, using items from the Self-Report Habit Index
Analysis A curve fitted to each person’s scores; “habit formed” defined as reaching 95% of that person’s plateau
Result Median 66 days to the plateau, range 18 to 254 days

The pattern behind the numbers is the study’s most useful finding. Automaticity rose quickly at first, then each additional repetition added less, until the behavior leveled off at a plateau. A habit does not switch on after a fixed number of days; it strengthens along a curve that flattens out.

What the Headlines Leave Out

The 66-day median describes a specific subgroup, and its limits change how it should be read.

  • Not everyone formed a habit. Of the 96 volunteers, 82 provided enough data. The researchers could fit the curve for 62 of them, and it fit well for 39. The 66-day median and the 18-to-254 range come from that best-fitting group. According to the British Psychological Society’s summary of the study, about half of the participants had not reached a level of automaticity high enough to count as a habit by the end.
  • 254 days was never observed. The study lasted 84 days. Every estimate beyond that point, including the upper end of the range, is a projection from the fitted curve rather than a measured outcome.
  • The 95% threshold is a convention. Choosing a different cutoff would shift every number. It is a way to compare people, not a moment when a habit begins.
  • Automaticity was self-reported. Participants rated how automatic the behavior felt. The authors acknowledged that this depends on people’s ability to judge their own behavior accurately.
  • The behaviors were simple. Drinking water or eating fruit with a meal is not the same as building a writing routine or quitting smoking, and a sample recruited at one university cannot represent every age group.

None of this makes the study weak; it was designed as a first look at how habits form in real life, and its authors were careful about these caveats. The problem is the popular version, which turns a median from one subgroup into a universal deadline.

Why the Range Is So Wide

Lally’s team found that the type of behavior mattered. Participants who chose an exercise behavior took about one and a half times as long to reach their plateau as those who chose an eating or drinking behavior. Consistency mattered too: people who performed the behavior more regularly showed a clearer habit-formation curve.

Later research confirmed that the behavior itself sets much of the timeline. In 2023, Anastasia Buyalskaya, Colin Camerer and colleagues published in PNAS an analysis that skipped self-reports entirely. Using machine learning on objective records, more than 12 million observations of gym attendance from over 30,000 gym members and more than 40 million handwashing observations from over 3,000 hospital workers, they measured when each person’s behavior became highly predictable. Gym habits typically took months to form; Caltech’s summary put the average at about six months. Handwashing on hospital shifts became habitual within weeks.

Study Behaviors How habit was measured Time to habit
Lally et al., 2010 (n=96) Simple eating, drinking and activity behaviors Self-rated automaticity Median 66 days; range 18 to 254 days
Keller et al., 2021 (n=192) Everyday nutrition behaviors Self-rated automaticity Median 59 days among those who formed a habit
Buyalskaya et al., 2023 (30,000+ gym members, 3,000+ hospital workers) Gym attendance and handwashing Predictability of recorded behavior Gym: months (about six on average); handwashing: weeks

A systematic review by Ben Singh and colleagues at the University of South Australia (Healthcare, 2024) pulled together 20 studies with 2,601 participants. Only four of them measured how long habit formation took. Those reported medians of 59 to 66 days and means of 106 to 154 days, with individual estimates from 4 to 335 days. The gap between median and mean is telling: a few slow habit-formers pull the average far above the typical case. The reviewers also rated most included studies as having a high risk of bias, a reminder that this research field is still young.

What a Missed Day Does

One of the most practical findings in Lally’s study is often overlooked. Missing a single opportunity to perform the behavior did not materially affect the habit-formation process. The automaticity built up so far did not reset after one lapse. What did hurt was repeated inconsistency: participants who performed the behavior irregularly showed weaker habit formation overall.

That distinction matters for anyone tracking a streak. An unbroken chain of check marks feels motivating, but the evidence suggests the chain is not what builds the habit. Repetition in a stable context is. A single missed day is a minor dip in the curve, not a return to day one.

What Speeds Habit Formation

The studies point to a handful of factors that shorten the road. Singh and colleagues highlighted how often a behavior is performed, when it is practiced, and whether people enjoy it, and the review reported stronger results for behaviors people chose themselves. Combined with the earlier findings, the evidence supports a few concrete steps:

  • Tie the behavior to a fixed cue. Every participant in Lally’s study linked the new behavior to an existing part of the day, such as “after breakfast.” The cue is what eventually triggers the action without deliberation.
  • Repeat it daily if possible. Daily behaviors give the habit-forming curve more repetitions per week; behaviors done less often, like weekly gym sessions, take longer in calendar time.
  • Start with the easier version. In Lally’s study, eating and drinking behaviors reached their plateau faster than exercise. A ten-minute walk after lunch is closer to those quick-forming behaviors than a full workout program.
  • Plan for months, not weeks. People routinely underestimate how long personal projects will take, a bias explored in the planning fallacy. Expecting a habit to settle in 21 days sets up a sense of failure at exactly the point where progress is still being made.

For work habits, the same logic applies to routines such as starting the day with one focused block before checking email. A fixed trigger, like opening the laptop, followed by a short, repeatable session, for example a single interval of the Pomodoro technique, has the features that made habits form faster in the research: a stable cue, daily repetition and a simple action. That is an application of the findings, not something these studies tested directly.

The Calendar Math for a New Year’s Habit

The difference between 21 and 66 days is easier to feel on a calendar. For a habit started on Friday, January 1, 2027, counting January 1 as day 1:

Milestone Day number Date in 2027
Fastest habit in Lally’s study Day 18 Monday, January 18
The 21-day myth Day 21 Thursday, January 21
Median in Lally’s study Day 66 Sunday, March 7
Slowest projected habit in Lally’s study Day 254 Saturday, September 11
Slowest estimate in the 2024 review Day 335 Wednesday, December 1

On these figures, a typical new habit would not settle until early March, and a slow one could still be forming in the fall. Converting a target day into a date works the same way as finding the day of the year, counted forward from the start date. The 21-day mark falls in the third week of January, when only the very fastest habit in Lally’s study had reached its plateau and the typical one was still on the steep, early part of its curve.

 


Sources and references

Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W. and Wardle, J. (2010): How Are Habits Formed: Modelling Habit Formation in the Real World, European Journal of Social Psychology
The original study behind the 66-day median and the 18 to 254 day range
Singh, B., Murphy, A., Maher, C. and Smith, A. E. (2024): Time to Form a Habit: A Systematic Review and Meta-Analysis of Health Behaviour Habit Formation and Its Determinants, Healthcare
Review of 20 studies with 2,601 participants, including all studies that measured time to habit formation
Buyalskaya, A., Ho, H., Milkman, K. L., Li, X., Duckworth, A. L. and Camerer, C. (2023): What Can Machine Learning Teach Us About Habit Formation? Evidence From Exercise and Hygiene, PNAS
Analysis of more than 12 million gym attendance observations and 40 million handwashing observations using objective records
Gardner, B. (2012): Busting the 21 Days Habit Formation Myth, UCL Health Behaviour Research Blog
How the 21-day claim was traced to Maxwell Maltz’s 1960 book Psycho-Cybernetics
↑
✕Close Menu