How long does it take to break a habit? The number you were hoping for doesn't exist, and the reason it doesn't is more useful than the number would have been. Every popular figure — 21 days, 30 days, 66 days — was measured on people adding a behaviour, not removing one. Breaking works differently enough that borrowing the number quietly sets you up to quit in week four, on schedule, feeling like you failed.
What actually sets the clock is how often the cue fires. Not how long you've been trying. A habit whose trigger appears thirty times a day and one whose trigger appears twice a week are two completely different projects, and the second one takes longer on the calendar even though it costs far less effort per day.
Where the 21-day and 66-day numbers came from
The 21-day figure traces back to Maxwell Maltz, a plastic surgeon who noticed in the 1960s that patients took roughly three weeks to stop seeing a stranger in the mirror. That's an observation about self-image, made about surgery. It was never a habit study, and it has been repeated so often that it now sounds like one.
The 66-day figure is real research and still doesn't answer this question. The UCL work found automaticity forming anywhere between 18 and 254 days, with 66 as an average, and it measured people building something new: a glass of water with lunch, a walk after dinner. Nobody in that study was quitting anything.
If it's the building side you're after, why streaks break covers that spread and what it means for a new habit. The rest of this post is about the other direction.
You aren't deleting a behaviour, you're outcompeting a cue
The mechanical difference matters. When you build a habit you're writing a new cue-response link, and repetition strengthens it. When you break one, the old link doesn't get erased. It stays there, intact, and what you're doing is teaching a second response to the same cue and making that one win more often.
That's why someone can quit smoking for eleven years, walk into a specific pub they used to smoke outside, and feel the pull immediately. The learning didn't vanish. The context found it. Habit loop examples pulls this apart on five ordinary behaviours, and the practical consequence is blunt: "how long until it's gone" is the wrong question. "How long until the new response wins reliably" is the answerable one.
Count cue exposures, not days
If the new response wins by being practised, then the unit of progress is a cue exposure — a moment where the trigger fires and you do something else instead. Days are just a proxy, and a bad one.
Work out roughly how often your cue fires, and the calendar estimate falls out of it:
| Habit you're breaking | How often the cue fires | Practice chances per week | Realistic calendar time |
|---|---|---|---|
| Nail biting | ~30 a day, mostly unnoticed | ~210 | 4–8 weeks |
| Checking the phone on waking | Once a day, same moment | 7 | 2–4 months |
| The after-dinner cigarette | Once a day, highly predictable | 7 | 3–6 months |
| Doomscrolling on the sofa | 2–3 an evening | ~15 | 6–10 weeks |
| Ordering takeaway on a bad day | Twice a week, mood-driven | 2 | 6–12 months |
| Snapping at someone when tired | 3–4 a week, unpredictable | 3 | 6+ months |
These are illustrations, not findings — no study has measured breaking times per cue exposure the way the UCL work measured building times per day. But the shape holds up against how people actually experience quitting, and it inverts the usual intuition. The thirty-times-a-day habit feels like the monstrous one because it's exhausting today. It's the fast one. Two hundred practice attempts arrive inside a fortnight, and by week six the new response is winning most of them.
The genuinely slow ones are the rare-cue habits. If your trigger is "a bad day at work", you get maybe two attempts a week, so eight months of real progress looks, from the inside, like nothing happening. People quit rare-cue habits not because they're harder but because the feedback is too sparse to feel like feedback.
The catch rate is the number that moves first
There's a second variable hiding in that table, and watching it is what keeps a rare-cue habit alive.
Early on, most cue exposures go unnoticed. Your hand is at your mouth before you register the urge. What improves first isn't your resistance — it's your catch rate, the fraction of exposures you notice in time to choose. It typically goes something like: week one you catch one in five, week three one in two, week six four in five. Resistance barely changes. Noticing changes enormously.
That matters because catch rate is measurable and encouraging in a period when the clean-day count is still embarrassing. Someone on day nine with three slips has, on the arithmetic above, probably tripled their catch rate — which is real progress that a streak counter renders as "3 days".
The slip is in the timeline, not outside it
Every realistic estimate above assumes slips. Plan for them, because the thing that actually extends a break from three months to never is the reaction, not the lapse.
Psychologists call it the abstinence violation effect: one slip gets read as proof the whole attempt has failed, so you abandon it, and the abandonment costs far more than the cigarette did. A single slip in six weeks moves your underlying learning by almost nothing. Two weeks of "well, I've blown it anyway" moves it a lot.
So separate the two numbers in your head. The clean-day count resets. The learning doesn't. If your count is back at zero for the third time but your catch rate is climbing, you're further along than you were on your best previous run, and the count is simply not the thing that was improving. Getting back on track is the more useful skill here than any amount of resolve — and a tracker pointed at breaking habits covers the count-the-clean-day mechanics this post deliberately skips.
How long does it take to break a habit: numbers to plan with
Something like this: high-frequency habits, four to eight weeks to the point where the new response usually wins, with slips throughout. Once-a-day habits, two to six months. Mood-triggered habits with rare cues, most of a year, and you will need something other than a streak to keep you interested during it.
None of those are promises. They're planning numbers, and having a realistic one beats having 21 days, because 21 days guarantees a moment three weeks in where you conclude it isn't working. It is working. It's just that on a once-a-day cue you've had twenty-one attempts, and twenty-one attempts is not many. Examples of bad habits lists the common ones with the replacement that tends to fit each cue, which is the other half of the job.
FAQ
How long does it take to break a bad habit?
There's no single number, and the realistic range runs from about a month to most of a year. What decides it is how often the cue fires: a habit triggered thirty times a day gives you hundreds of practice attempts inside a fortnight, while one triggered by a bad week gives you two a month and takes correspondingly longer on the calendar.
Is the 21-day rule for breaking habits true?
No. It comes from a 1960s plastic surgeon's observation about how long patients took to adjust to their new appearance, not from research on habits. The best-known actual habit study measured 18 to 254 days for a new behaviour to become automatic, and it wasn't measuring breaking at all.
Does a habit ever go away completely?
Not exactly. The old cue-response link stays available, which is why a familiar place or mood can produce a strong pull years later. What changes is which response wins by default, and after enough practice the new one wins so consistently that the old one stops mattering day to day.
Does tracking help you break a habit faster?
It helps mainly by making sparse progress visible and by showing you when your cue actually fires. init.Habits is a habit tracker for iPhone with a terminal-style interface, per-habit statistics, earned streak freezes, GitHub-style heatmaps, a pomodoro timer and 23 editor themes — the freezes matter here, because one slip costing a banked day instead of the whole count is what stops a lapse from turning into a relapse.