How long does it really take to form a habit?

You started a new habit three weeks ago. You were told it would be automatic by now. It isn’t. Every morning still feels like a decision.

That doesn’t mean you’re bad at habits. It probably means the number you were given was wrong.

Where “21 days” comes from

The 21-day idea is usually traced to the book Psycho-Cybernetics by Maxwell Maltz, not to a study. A 2024 review of the research says this popular timeline has been questioned and that the evidence does not support it (Singh et al., 2024).

What the research says

The best-known study. Phillippa Lally and colleagues asked 96 volunteers to pick one healthy eating, drinking or exercise habit. They were university students recruited by email circular, mostly postgraduates (mean age 27). Each person did the habit once a day, in the same situation (for example “after breakfast”), for 12 weeks, and every day they rated how automatic it felt (Lally et al., 2010).

Among the 39 people whose data fitted the researchers’ model well, the median time to reach 95% of their automaticity plateau was 66 days, with a huge range: 18 to 254 days. Some habits settled in under three weeks. Others were still getting stronger long after the study ended.

A 2024 review. Ben Singh and colleagues pooled 20 studies with 2,601 people. Only four of them measured how long a habit took. Those four reported medians of 59 to 66 days and mean times of 106 and 154 days (both from one study of morning versus evening stretching), with individual times from 4 to 335 days. The authors suggest expecting at least two to five months for a health habit. They also note that 11 of the 20 studies were rated high risk of bias, so the numbers are a rough guide, not a rule (Singh et al., 2024).

A few details that help. In the review, habits practised in the morning and habits people chose themselves tended to be stronger. In Lally’s study, the early repetitions added the most automaticity, and gains levelled off later.

What about a missed day?

This is the part most people worry about. Lally’s team looked at 140 missed days. Missing once caused only a very small decrease in how automatic the habit felt (0.29 points on a 0 to 42 scale), and the authors found no longer-term cost from a single omission. The study only tested single missed days, so one miss does not reset you to zero; what a week of misses does was not measured here (Lally et al., 2010). There’s more on this in a missed day is data, not defeat.

Why MileStone is built this way

If a habit can take two months or more to feel automatic, the long middle has to feel worth it. That shaped three choices.

• A habit can count toward a goal. In MileStone you can link a habit to the goal it serves, and the time you put in counts toward that goal’s pace. A morning run is hours toward the race, not just a tick. See how habits link to goals.
• A missed day shows 💤, not a red cross. Because one miss had no lasting cost in the research, we didn’t want the app to treat it as a failure.
• Consistency over the month, not just the streak. The app shows things like “26 of 30 days this month”, so one missed day doesn’t wipe out the other 26. If a routine is still a struggle after a month, MileStone shows which steps you really do and offers to park the rest.

None of this makes a habit form faster. It is meant to keep the numbers honest while it does. The app is at codemilestone.com/app.

How to apply it

1. Plan for months, not weeks. Put a review date about 10 weeks out, not 3.
2. Tie the habit to the same moment each day, like “after I make coffee”.
3. Pick a habit you actually chose, and start smaller than feels impressive.
4. If you miss a day, do the next one. Watch for patterns, not single misses.
5. Track it somewhere you’ll see it, on paper or in any app.

If the missed day is what makes you want to quit, read why a missed day is data, not defeat.