Sixty-six days. That is the average it actually took a group of volunteers at University College London to turn a new behavior into something close to automatic. Not three weeks. Not 21 days. Sixty-six, with a spread underneath it that runs from eighteen days to two hundred and fifty-four.

The 21-day figure most of us have absorbed comes from somewhere else entirely — a plastic surgeon’s offhand observation in a 1960 self-help book. I have read it in productivity posts, and quietly half-believed it myself for years. When I finally went looking for the source, the trail led to a clinician, not a study.

A quick note before going further: I am a curious generalist who reads behavior-change books and tries them out, not a psychologist or any kind of clinician. What follows is reading and reflection on one piece of research, and the study here is a finding from a particular group of people, not a settled law about everyone. Treat it as a clue, not a prescription.

Where the 21-day idea came from

The number traces to Maxwell Maltz, a plastic surgeon whose 1960 self-help book Psycho-Cybernetics seeded the whole thing. As Scientific American recounts, Maltz observed that it took his patients roughly 21 days to get used to their new appearance after surgery. From there he generalized, applying the same figure to all sorts of changes, from getting used to a new house to shifting a long-held belief.

Notice what that actually was: a clinician sharing an impression of how long his patients seemed to take to adjust, not a measurement of anything. No formal experiment was run to verify it. Honest enough as an observation, but not a measurement.

What happened next is the familiar story of how a soft observation hardens into fact. Over decades of self-help repetition, Maltz’s loose “about 21 days to adjust to a new face” lost its hedges, lost its surgical context, and became the crisp, confident “it takes 21 days to form a habit,” with the qualifier dropped and the number kept.

What the UCL research found

A real-world measurement came much later. Phillippa Lally and colleagues at University College London asked 96 volunteers to pick a daily eating, drinking, or activity behavior tied to a consistent cue, then tracked how automatic it felt over twelve weeks. Real people, real behaviors, measured over time rather than guessed at.

The headline number was not 21. Lally reported that “the average time (among those for whom our model was a good fit) was 66 days.” Hold on to that parenthetical, because it matters. Sixty-six days was an average for the subset of participants whose data fit the model well, not a fixed deadline that applies to any habit you might try to build.

The spread underneath that average is the part I find most useful. The time to reach near-automaticity ranged from 18 to 254 days across participants. Some participants got there in eighteen days; others took two hundred and fifty-four. That is not a rounding error — it is the difference between a few weeks and the better part of a year, depending on the person and the behavior.

So the real answer to “how long does it take” is something closer to “it depends, enormously” — less tidy than 21, and more honest.

The finding that might change how you approach habits

The piece I keep coming back to is this: Lally and her colleagues found that a single missed day did not undo the work. The chain did not shatter because of one slip.

Lally put it this way: “In our study we showed that missing one opportunity did not significantly impact the habit formation process, but people who were very inconsistent in performing the behaviour did not succeed in making habits.” Consistency over the long run still mattered. A pattern of skipping was a problem. A single off day was not.

Repeat the behavior every day if you can, but don’t write the whole thing off if you miss a day or two. That is the reading I think is right.

It matches what I have learned the slow way. I like routine, but I have been honest enough with myself to admit I cannot always hold one. The routine I run now is the latest iteration of many, arrived at by trial and error rather than by some clean, unbroken streak. Short breaks from it are fine. Past a certain point the wheels do come off, and I am not going to pretend otherwise.

The 21-day framing can make every missed day feel like a reset to zero, like we have snapped the chain and have to start the count again. The UCL finding says that is not how it works. The whole thing is messier and more forgiving than the myth lets on.

Somewhere between eighteen days and two hundred and fifty-four, for reasons that have to do with the person and the behavior and the day, a thing you have been doing becomes a thing you do. That is most of what the research seems to say. The rest is what each of us makes of it.