The "21 days to form a habit" rule is wrong — University of South Australia research found habits begin forming around 59 days in, and can take up to 335 days, meaning most people quit just as the process is finally getting started

The promise behind a 21-day habit challenge is clean and reassuring: repeat a behaviour for three weeks, reach the finish line, and watch it become automatic. The evidence is much less tidy.

A University of South Australia-led review found that the few studies directly estimating habit-formation time reported medians of 59 to 66 days, means of 106 to 154 days, and individual estimates ranging from four to 335 days. Those figures make a three-week rule difficult to defend.

Two qualifications matter immediately. Habit strength generally builds from the first repetitions rather than suddenly beginning on day 59. The review also did not measure when most people abandon a new routine, so it cannot show that most quit at the exact moment the process starts working.

The finding is worth taking seriously, but it should not be read as the final word. This was one systematic review, and only four of its 20 included studies directly reported how long habit formation took.

What the University of South Australia review examined

Ben Singh and colleagues searched six academic databases for experimental studies of health-related habits. Their review, published in the journal Healthcare in December 2024, included 20 studies and 2,601 participants whose reported mean ages ranged from 21.5 to 73.5 years.

The behaviours were specific. They included physical activity, drinking water, eating patterns, dental flossing, daily stretching, reducing sedentary time, taking vitamins and microwaving a dishcloth as a food-safety practice. This was not a review of every habit people might want to form, from writing each morning to checking email less often.

Most studies measured habit strength using self-report questionnaires. These asked, in different forms, whether a behaviour felt automatic, required thought or belonged to a routine. The meta-analysis found that habit scores tended to rise between the start and end of the interventions, but timing evidence was sparse. Eleven of the 20 studies received a high risk-of-bias rating from the reviewers.

That combination is useful but limited: a broad synthesis of health-behaviour studies resting on a small timing subset.

Day 59 was a threshold, not a starting gun

The most eye-catching range came from a randomised trial led by Jessica Keller. In that study, 192 adults selected an everyday nutrition behaviour and connected it either to a time, such as 9 am, or to an existing routine, such as breakfast. Participants completed daily questionnaires for 84 days.

Among those who successfully formed a habit, the modelled time to reach 95 per cent of peak automaticity ranged from four to 335 days, with a median of 59 days. Only 23 per cent of the participants reached the study’s predetermined habit-formation threshold.

The 335-day endpoint therefore does not mean every habit requires nearly a year. It was the far end of a wide individual range produced by a model, while 59 days described the midpoint among successful cases in that particular study.

Nor were participants doing nothing psychologically until the threshold arrived. Habit formation was modelled as a rising curve. Repeating a behaviour in a recurring context gradually strengthened the association; gains then slowed as automaticity approached a personal plateau. Day 59 was a convenient summary of that curve, not the first day on which learning occurred.

Why 21 days became more persuasive than the evidence

The 21-day idea is commonly traced to plastic surgeon Maxwell Maltz, who wrote in 1960 about the minimum period his patients seemed to need to adjust to changes in their bodies and self-image. A personal observation about adjustment was gradually converted into a universal formula for forming any habit.

Later research did not find a replacement magic number. In a 2010 study published in the European Journal of Social Psychology, Phillippa Lally and colleagues followed volunteers who chose an eating, drinking or activity behaviour to repeat in the same context over 12 weeks. For participants whose automaticity followed the model well, the estimated time to reach 95 per cent of its plateau ranged from 18 to 254 days. The often quoted midpoint was 66 days.

One useful detail from that study was that missing a single opportunity did not materially derail the formation curve. The research did not treat consistency as a streak that instantly reset to zero after one lapse.

Replacing 21 with 59 or 66 would recreate the same mistake with a better dataset. The central result across these studies is variation, not a new universal deadline.

The review did not find when most people quit

The University of South Australia paper did not track a general population from New Year’s resolution to abandonment. It did not estimate the proportion of people who stop after 21 days, and it did not compare quitting dates with the point at which automaticity begins to grow.

The authors did argue that short challenges are often too brief for habits to become established, particularly for more complex behaviours. That is a reasonable interpretation of the reported timelines. It remains different from demonstrating that most people quit just as formation starts.

What a rigid 21-day promise can create is an expectation problem. Someone may repeat a behaviour for three weeks, notice that it still requires conscious effort, and conclude that the method or the person has failed. The studies suggest that requiring thought after 21 days is entirely compatible with an ordinary formation process.

I have written before about why consistency is often better understood as a system than a fixed supply of willpower. The newer review adds a sober timescale to that idea, without turning it into a guarantee.

Different behaviours produced different timelines

The review associated stronger habits with greater repetition, stable timing or context, and behaviours people had chosen themselves. Morning practices tended to show greater strength than evening ones in the included studies. Simpler practices, such as drinking water, may also become automatic differently from activities that require more planning and effort.

These were patterns across a mixed and limited evidence base, not instructions that work equally for everyone. The review’s authors called for larger and better-designed studies, with more direct measurement of the time needed for different habits.

The research leaves no defensible countdown that applies to every person and behaviour. It replaces a three-week promise with a more modest observation: automaticity usually develops gradually, and the calendar can vary by months even when two people are trying to establish something that looks similar.