There is no scientifically supported rule that a habit becomes automatic after exactly 21 days.

A 2024 systematic review and meta-analysis examined 20 habit-formation studies involving 2,601 participants. In the four studies that reported a time to habit formation, median times were 59 to 66 days, mean times were 106 to 154 days, and individual results ranged from 4 to 335 days. The authors also noted that 11 of the 20 included studies had a high risk of bias. Singh et al., 2024

That does not mean a habit “takes 66 days” either.

The research points to a more useful conclusion: habit formation varies widely between people, behaviours and contexts, and many health habits take months rather than a fixed number of days to become more automatic.

What does it mean for a behaviour to become a habit?

In habit research, a habit is not simply something you have done several times.

The central idea is automaticity. Repeating a behaviour in a consistent context can gradually make that context trigger the behaviour with less conscious effort.

One influential real-world study by Phillippa Lally and colleagues followed 96 volunteers who chose an eating, drinking or activity behaviour to perform daily in the same context for 12 weeks. Participants repeatedly reported whether they performed the behaviour and how automatic it felt. Lally et al., 2010

Among participants for whom the researchers could model the process, time to reach 95% of the estimated automaticity plateau ranged from 18 to 254 days. The increase followed a curved pattern: automaticity rose more quickly at first and then increased more slowly as it approached a plateau.

This is very different from a switch flipping on a predetermined day.

Where does the “66 days” number fit?

The number is useful only if its limitations are understood.

The 2024 systematic review found that studies reporting median habit-formation times produced values of 59 to 66 days. But mean values in other studies were much longer, from 106 to 154 days, and individual results ranged from 4 to 335 days. Singh et al., 2024

So 66 days is not a deadline and not a guarantee.

It is better interpreted as one estimate from a research literature showing substantial variability.

If a new routine still requires effort after two months, that does not mean it has failed.

Do simple habits form faster than difficult ones?

The evidence suggests that the type and complexity of the behaviour matter.

The 2024 systematic review identified behaviour type, frequency, timing, personal choice, affective judgments, behavioural regulation and preparatory habits among factors associated with habit strength. It also reported substantial variation between behaviours and individuals. Singh et al., 2024

That makes intuitive comparisons dangerous. Drinking a glass of water after breakfast is not the same task as completing a 45-minute workout, preparing a complex meal or going to bed an hour earlier.

The review does not provide a formula that predicts how quickly one specific person will form one specific habit. What it does show is that habit timing varies substantially across behaviours and individuals, so one universal number is not supported.

Does missing one day ruin the habit?

The original Lally study provides reassuring evidence on this point.

Missing one opportunity to perform the chosen behaviour did not materially affect the habit-formation process in that study. Lally et al., 2010

That does not mean consistency is irrelevant. Repetition was central to the study, and more consistent performance was associated with a better fit to the habit-formation model.

The useful distinction is between an occasional miss and repeated non-performance.

One missed day is not evidence that you have returned to zero.

Is it better to use a time cue or a routine cue?

Both can work.

A 2021 randomized controlled trial assigned 192 adults to form a daily nutrition habit using either a time-based cue or a routine-based cue. For example, one person might link the behaviour to a specific clock time while another links it to an existing routine. Participants reported their behaviour and automaticity for 84 days. Keller et al., 2021

Among participants who successfully formed habits, the median time to peak automaticity was 59 days. The study found no significant difference between the two cue types. Repeated enactment of the planned behaviour was a key predictor of automaticity.

The practical lesson is not that 8:00 a.m. is better than “after breakfast” or vice versa.

It is that a cue needs to be clear enough to support repeated performance.

What makes a useful cue?

A cue should happen reliably and make the desired behaviour easy to identify.

Examples include:

after breakfast;
when you start work;
after brushing your teeth;
when you return from lunch;
at a specific time when your schedule is usually stable.

For drinking water, a cue such as “after breakfast” may be easier to follow than the vague intention “I should drink more today.”

A reminder can also function as a cue, but the reminder itself is not the habit. The repeated behaviour in response to the cue is what matters.

The studies above do not establish an ideal number of phone notifications. They support the broader point that the behaviour has to be repeatedly performed in relation to a cue if automaticity is to develop.

Do reminders create habits?

Research supports reminders as one possible way to cue behaviour, but reminders should not be confused with automaticity.

The 2021 cue-planning trial showed that both routine and time cues can support habit formation when the planned behaviour is repeatedly enacted. Keller et al., 2021

A phone notification can create a stable time cue, but if the action is rarely performed, the cue alone is not enough.

For a hydration routine, a reminder can be used as a cue, but the evidence does not prescribe a universal reminder schedule. A simple implementation is to choose a time or routine cue that occurs reliably, make water available at that point, and observe whether the drinking behaviour is actually repeated.

Our guide How to Remember to Drink Water applies these principles specifically to hydration.

Is a streak the same as a habit?

No.

A streak measures consecutive completion. Habit research is interested in how automatic a behaviour becomes in response to a context.

Those concepts can overlap, but they are not identical.

A person can maintain a streak through effort, planning and motivation while the behaviour still feels deliberate. Another person may miss one day but still have a strong automatic tendency to perform the behaviour in the usual context.

The Lally study is particularly relevant here because a single missed opportunity did not materially disrupt the modeled habit-formation process. Lally et al., 2010

A streak should not be treated as a measurement of automaticity.

What does the newest systematic review tell us?

The 2024 review is useful because it combines a broader group of health behaviours than any single experiment.

It included studies on:

physical activity;
drinking water;
diet;
flossing;
vitamin consumption;
reducing sedentary behaviour;
other health-related actions.

Across the included interventions, habit scores improved significantly from before to after the interventions. But the authors emphasized that the evidence base is still limited. Only four studies reported actual time-to-habit values, and most included studies had a high risk of bias. Singh et al., 2024

This is exactly why precise claims such as “your habit will form in 21 days” or even “your habit will form in 66 days” go beyond the evidence.

A research-based way to build a hydration habit

Drinking water is one of the health behaviours included in habit-formation research, but the same principles apply as with other simple behaviours.

One way to apply those findings to hydration, without claiming that the studies validated a specific water-reminder protocol, is:

Choose one small behaviour. For example, drink a glass of water after breakfast.

Choose one stable cue. Keep the cue the same on most days.

Make the behaviour possible at the cue. For example, have water available where the chosen routine occurs.

Repeat the same behaviour in the same context. Consistent repetition is the mechanism examined in the cited habit studies.

Expect variation. Automaticity may develop slowly, and the Lally study found that one missed opportunity did not materially disrupt the formation process.

Review what is happening. Ask whether the behaviour is beginning to occur with less deliberate thought, which is closer to how habit research defines automaticity than simply counting a streak.

This approach does not promise a habit by a particular date. It matches the evidence better than a fixed-day challenge.

What if the habit still feels difficult after two months?

That is not unusual.

The systematic review found mean formation times beyond three months in some studies and individual values extending much longer. Singh et al., 2024

A slow result is therefore compatible with the published range and is not, by itself, evidence that the process has failed. The studies do not provide a validated troubleshooting checklist for every individual, so changing a cue or behaviour should be treated as a practical experiment rather than a guaranteed fix.

What the research leaves us with

Habit formation does not run on a universal countdown.

The best recent systematic review found median values of 59 to 66 days in studies that reported them, means of 106 to 154 days, and individual times ranging from 4 to 335 days. The evidence base is still limited, and many included studies had a high risk of bias. Singh et al., 2024

Earlier real-world and randomized research supports the same broader idea: consistent repetition in a stable context can increase automaticity, but the speed varies substantially between people and behaviours. Lally et al., 2010 Keller et al., 2021

So do not build your plan around day 21 or day 66.

Build it around a clearly defined behaviour, a stable cue and repeated performance rather than a promised completion date.

Sources and further reading

This article discusses behavioural research and is not a substitute for medical or psychological care when a behaviour is connected to a health condition.