Three weeks was never the number.
Forget the '21 days' myth. A field study tracking real habits found the average was closer to 66 days — and the range was enormous.
Study title: How Long a Habit Actually Takes to Form — Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W., & Wardle, J., European Journal of Social Psychology, 2010
- Study type
- Longitudinal study
- Population
- 96 adult volunteers in the UK, each tracking one self-chosen daily eating, drinking or activity behaviour over 12 weeks
- Published
- 2010
- Evidence confidence
- Moderate confidence
The short version
Ninety-six people repeated one everyday behaviour daily for twelve weeks while rating how automatic it felt. On average, behaviours took 66 days to reach peak automaticity — with individual estimates ranging from 18 to 254 days. Missing the odd day did not meaningfully harm the process.
The finding, drawn out
66 days — the average
18–254 days · the full range across individuals
Average and range as reported by Lally et al. (European Journal of Social Psychology, 2010).
Why it matters
A new routine that still feels effortful after a month isn't failing. It is, in this data, entirely ordinary — and one missed day is not the end of it.
What this doesn't mean
It does not mean everyone needs 66 days. That is an average across a small, motivated, self-selected sample doing simple behaviours; the spread between people was enormous.
01
The question
02
How it was studied
03
What they found
04
Limitations
05
What it may mean
In real life
Someone starting a new evening walk might expect it to feel automatic within a month. This research suggests two or three months is a more realistic timeline, and that occasionally skipping a day is unlikely to derail progress.
Try this
Pick one small, simple behaviour to repeat in a fixed context (same time, same trigger) and track it for at least eight weeks before judging whether it's 'sticking' — that's closer to the real timeline than a few weeks.
Evidence confidence
Moderate confidence
A single, frequently cited field study with a real-world design, but a small, self-selected sample and self-reported outcomes; later attempts to replicate the exact timeline have found similar wide variability rather than a fixed number.
Source
Lally, P., van Jaarsveld, C. H. M., Potts, H. W. W., & Wardle, J. (2010). How Long a Habit Actually Takes to Form. European Journal of Social Psychology.
Peer-reviewed · Free full text available · Last reviewed 2026-08-14
Read the original research →One study isn't the whole storyOne study gives us a piece of the picture. Selfey looks for patterns across the broader evidence rather than treating a single finding as a rule.
Read this in everyday terms
Your Environment Is Doing More Work Than Your WillpowerSelfey research briefs are educational summaries, not medical advice, diagnosis or treatment. Research describes patterns across groups; it cannot tell you what is right for your situation.
Related research
- LifeWhy 'If-Then' Plans Outperform Good IntentionsPairing a goal with a specific 'if this happens, then I'll do that' plan is associated with meaningfully better follow-through than intention alone.
- LifeIt's Not the Workload Alone — It's the Workload Without ControlHigh demands paired with low control are associated with more strain than high demands alone — a degree of decision latitude appears to matter as much as workload.
- LifePredictable Routines Help Kids Feel Less StressedPredictable daily routines are linked with lower stress and steadier behaviour in children, likely because less mental energy goes into figuring out what happens next.