Sticking with it

How long does the average diet last? What the research found.

About three to five weeks. Which means if you are somewhere around there and thinking of stopping, you are not weak. You are on schedule. That is the moment this page is written for.

What follows is what studies actually measured about when people stop and why, including the bit where three of the most repeated explanations turn out to have nothing behind them. Then six reasons to keep going that hold up better than wanting to look different.

Published 27 August 2026. This is a summary of published research, not medical or dietary advice.

When people actually stop

Towers and colleagues, in Public Health Nutrition in 2020, modelled fifteen years of US recipe searches and estimated that new January dieters lasted 3.12 weeks on one popular plan and 5.32 weeks on another. It is a modelled population average rather than an observed dropout curve, so it describes a window, not a person.

The prospective numbers agree. Norcross and Vangarelli followed 200 people making New Year’s resolutions for two years: 77% were still going at one week, 55% at one month, 43% at three months, and 19% at two years.

There is no cliff, and that matters

The best dataset on the shape of it is a study of 7,752 gym members whose attendance was tracked by turnstile for three years. Visits per month ran 5.51 in month two, 5.01 in month three, 4.61 in month four, 4.32 in month six, then flattened out. A smooth slide of roughly 8% a month, with no spike anywhere.

There is no cursed week to survive. Abandonment is a slope, not a cliff, which means there is no single day you have to get past, and no point at which it was ever going to suddenly get easy.

That is worse news and better news at once. Worse, because nobody is coming to rescue you at day 22. Better, because it means a bad week is not a wall you failed to climb. It is just the week you happened to be on the part of the curve where most people let go.

Three explanations that do not survive contact with the evidence

“Willpower runs out.” The idea that self-control is a finite resource that depletes through the day failed two of the largest replication efforts psychology has run. 23 laboratories and 2,141 people found an effect of 0.04, and 36 laboratories with 3,531 people found 0.06, with the data four times more likely under the null. Decision fatigue is a satisfying story about why you gave in at 9pm. It is not a finding.

“People set unrealistic goals and get disappointed.” Tested directly. Fabricatore and colleagues followed 180 people and found that having, and failing to meet. Unrealistic expectations predicted lower satisfaction but did not predict dropout or regain. The chain everybody assumes was checked, and it broke.

“90% quit the gym within three months.” This one has no primary source at all. The pages repeating it cite each other. When Sperandei and colleagues measured it at a Rio de Janeiro fitness centre in 2016 they found 47% gone by two months and 86% by six. Serious numbers, but not the famous one, which appears to have been invented.

What does predict quitting, in a study of 5,274 people on a commercial programme, is not seeing anything happen. Those whose weight had not moved by the end of month one were 1.86 times more likely to drop out early. Not disappointment in the abstract. The plain absence of a result.

When you feel like quitting, don’t

Here is the finding I would put on a wall. In that two-year study of 200 resolvers, 53% of people slipped at least once, and the average number of slips was fourteen. That figure includes the people who succeeded.

The people who got there were not the people who never slipped. They slipped roughly fourteen times and carried on anyway. From the inside, succeeding and failing look identical for the first fortnight. The only difference is what happens after the slip.

Which reframes the moment you are probably in. A bad day is not evidence that this is not working. It is the ordinary texture of the thing working. The decision that matters is not whether you eat the thing; it is whether the next meal goes back to normal or you write off the week.

And the background failure rate is nothing like the folklore. The “95% of diets fail” line traces to one 1959 study of 100 patients at a single clinic who were handed a diet with almost no support. Franz’s review of 80 randomised trials found people kept 3 to 6 kg off at up to four years, and no group returned to its starting weight. Not a triumph. But not the total collapse everyone has been told about either.

Six reasons that outlast wanting to look different

A systematic review of 66 studies found that the motivation you start with predicts the short term, and a different kind predicts whether you are still going a year later. So it is worth knowing which of these is actually yours.

Being able to do things with the people you love

Taking the sprint at sports day. Getting down onto the floor and back up again without thinking about it. Carrying someone you love when they are hurt.

None of that has been studied. Whether a fitter parent ends up playing more with their children does not appear to have been measured by anyone. The most common reason people give for getting in shape is also the least researched, which is worth admitting on a page that cites everything else. Nobody needs a trial to know whether they can still do those three things.

What has been measured is capability itself. In the PURE study of 139,691 adults across 17 countries, every 5 kg of lower grip strength was associated with a 16% higher risk of death from any cause. A stronger predictor in that dataset than systolic blood pressure. And in the LIFE trial, a randomised study of 1,635 sedentary adults aged 70 to 89, a structured activity programme reduced the incidence of major mobility disability from 35.5% to 30.1% over about two and a half years. That is an older population than most people reading this, and the trial is about disability rather than sports day. But it is the direction of travel, and it is causal evidence rather than a correlation.

Longevity, and the part of it you get a say in

You can still be hit by a bus. What the research offers is not control. It is a shift in the odds, and the size of it is worth knowing.

A pooled analysis of 654,827 people, published in PLoS Medicine in 2012, found that even half the recommended minimum activity. About 75 minutes of brisk walking a week. Was associated with 1.8 additional years of life after 40. At 22.5 MET hours a week the figure was 4.5 years. Being active and a normal weight, against inactive and obese, tracked 7.2 years.

Those are observational associations, not promises, and people who exercise differ from people who do not in ways no statistical adjustment fully removes. But 75 minutes of walking a week is a small ask attached to a large number, which makes it the most useful figure on this page.

There is also the smaller, nearer version of this. Turning up to a doctor’s appointment in decent shape is a different appointment from turning up ten kilos heavier. Not because you will be judged, but because more of what they find will be fixable.

Habits, and the point where it stops taking effort

Challenge yourself to get fit for a defined block of time and something useful happens: a bounded commitment with a start and an end is easier to hold than an open-ended intention to be better. The habits it builds are the part that outlives it.

Just do not expect them in three weeks. The 21-day figure comes from Maxwell Maltz’s 1960 book Psycho-Cybernetics, where it described how long plastic-surgery patients took to get used to their new face. It was a clinical observation about self-image, not a study of habits, and it has been quoted as science for sixty years.

When somebody finally measured it, Lally and colleagues found a median of 66 days for a new behaviour to become automatic, with a range from 18 to 254. A 2024 review of 20 studies and 2,601 people put the medians at 59 to 66 days, the means at 106 to 154, and the individual range at 4 to 335 days.

Learning what food actually does to you

Most people have never had a clean read on this, because they have never held anything steady long enough to notice. Log for a few weeks and the pattern shows up on its own: which meals leave you flat at three in the afternoon, what actually keeps you full, which “healthy” thing you eat daily is 600 calories.

On the measured side, the honest picture is narrower than the wellness version. Across 33 studies, each kilogram of weight lost was associated with a 0.13 mg/L fall in C-reactive protein, an inflammatory marker. In the IDEA trial. 454 adults with knee osteoarthritis, randomised, 18 months. Diet plus exercise produced about 11% weight loss and less knee pain than exercise alone. Note what that says: less pain. The best arm finished at 3.6 on a 0–20 scale, not zero, and the trial’s own imaging sub-study found no difference between groups in cartilage loss. Joints felt better; joints were not repaired.

Energy is the claim I would most like to make and can least support. The main meta-analysis found chronic exercise improved feelings of energy, and then reported that studies using placebo controls found no effect. Sleep is on firmer ground: across 66 studies, regular exercise had moderate benefits for sleep quality, with the largest gains among people sleeping worst to begin with.

How you see yourself

Something genuinely useful happens here, and it is not what people expect. In a meta-analysis of 57 exercise interventions, exercise produced a small improvement in body image compared with control conditions. The authors’ own word is “small”. More interesting is a study in which six 40-minute exercise sessions improved several aspects of body image while body weight and shape did not change at all. Small study, 34 people, but a clean result: what shifted was the perception, not the body.

Now the part that cuts against the obvious. If wanting to look different is your engine, the research says you have picked the weak one. Ingledew and Markland found appearance and weight motives fed externally-driven regulation and reduced participation, while health and enjoyment motives fed the kind that lasts. Segar’s year-long study of 156 midlife women found those with weight-loss goals did significantly less activity than those exercising for wellbeing and stress.

Which is the whole page in one finding: the reason that gets you started is measurably not the reason that keeps you going. Looking better is a real benefit. It is a bad engine. Let it be a side effect of the other five.

Sex, and what it does to a relationship

Nobody puts this on a fitness poster, and it has better evidence than most of what is on them. In a randomised trial published in JAMA in 2004, 110 obese men with erectile difficulties were assigned either to two years of weight loss and exercise or to control. Mean scores on the International Index of Erectile Function rose from 13.9 to 17.0 in the intervention group and did not move in controls; 17 men in the intervention arm reached a score in the normal range, against 3 controls. A later meta-analysis of seven trials and 478 men found exercise improved erectile function scores by 3.85 points on average.

Both studies were in men who already had difficulties, and the mechanism is not mysterious. Erectile function is largely vascular, so it tracks the same things your heart does. The evidence in women is much thinner and does not mirror it: reviews find pelvic floor training has the clearest effect, while general aerobic exercise gives inconsistent results. Presenting the two as symmetrical would be overstating it.

On relationships, one correction is worth having. The intuitive claim is that more sex makes couples happier. A 14-year study of 105 married couples found that sexual satisfaction tracked marital satisfaction while sexual frequency did not, once both were modelled together. And in research covering 30,645 people, the link between frequency and wellbeing levelled off at about once a week and held only for people in committed relationships. Correlational, both of them. Happier couples plausibly have more sex rather than the reverse.

Fuel8 is on the App Store.

A food diary for iPhone. It tracks protein, carbs and fat alongside calories, shows a weight trend rather than a daily verdict, never adds exercise calories back, and has no streaks to break.

Download on the App Store

Or leave your name and email for new recipes and what we add next. No more than that.

This form is the website, not the app. Your name and email go to Grow More Services L.L.C-FZ, the company that builds Fuel8, and are stored in Mailchimp so we can email you about new recipes and what we add next. This list is separate from your optional Fuel8 account. Unsubscribe in one click, any time.

If the number you set is the thing making this hard, that has its own research and its own page: what happens when you eat too little. And the walking arithmetic is over here.

Common questions

How long does the average diet last?

About three to five weeks. Towers and colleagues, publishing in Public Health Nutrition in 2020, modelled fifteen years of US recipe-search data and estimated new January dieters persisted 3.12 weeks on one popular diet and 5.32 weeks on another. That is a modelled population average from search behaviour rather than an observed dropout curve, so treat it as an estimate of the window rather than a measurement of any individual.

Is there a specific week when most people quit?

No, and the best data argues against it. DellaVigna and Malmendier tracked three years of turnstile records for 7,752 gym members and found attendance declined smoothly. 5.51 visits in month two, 5.01 in month three, 4.61 in month four, 4.32 in month six, with no spike at any particular point. Abandonment is a slope, not a cliff. Most people have stopped by around weeks four to five simply because that is where the curve falls steepest, not because anything happens that week.

Do 90% of people quit the gym within three months?

That figure has no traceable primary source. The pages repeating it cite each other. Sperandei and colleagues measured it properly in 2016 at a Rio de Janeiro fitness centre and found 47% had dropped out by two months and 86% by six months. So the real numbers are serious, but the famous one is folklore.

How long does it actually take to form a habit?

Longer than 21 days. That figure comes from Maxwell Maltz's 1960 book Psycho-Cybernetics, describing how long plastic-surgery patients took to adjust to their new appearance. It was never a study of habits. Lally and colleagues measured it in 2010 and found a median of 66 days, ranging from 18 to 254. A 2024 systematic review of 20 studies and 2,601 people reported medians of 59 to 66 days, means of 106 to 154, and individual times from 4 to 335 days.

Does slipping mean I have failed?

The research suggests the opposite is normal. In Norcross and Vangarelli's two-year study of 200 people making New Year's resolutions, 53% of participants slipped at least once, with an average of fourteen slips, and that includes the people who succeeded. Slipping was what success looked like from the inside, not a departure from it.

Is it true that 95% of diets fail?

That statistic traces to a single 1959 study of 100 patients at one clinic, given a diet with almost no support or follow-up. Franz and colleagues' 2007 review of 80 randomised trials found people lost 5-8.5 kg in the first six months and kept 3-6 kg off at up to four years, and no group returned to its starting weight.

If it has stopped feeling optional

Persisting is good advice right up until the point where counting is running the day rather than helping it. If that is where you are, none of the above applies to you and a target is not what you need. In the UK, Beat runs a helpline; in the US, ANAD. Both are worth a call earlier than you think.

Sources

This page summarises published research and is not medical or dietary advice. Fuel8 is a food-logging tool, not a medical device. Studies report group averages and individual results vary. If you have a medical condition, are pregnant, or have any history of disordered eating, speak to a qualified professional before changing how you eat or exercise. The full health disclaimer is below.