You have probably been told that 95% of people who lose weight regain it. That figure is not from a modern study. It comes from 100 patients treated at a clinic in 1959, where two years later only 2% had kept off 20 pounds or more. One small study, from the Eisenhower administration, repeated for sixty-six years.
So the pessimism is unfounded. But here is the part that usually gets left out of articles like this one: the optimistic figure most often used to replace it is built on sand too.
The reassuring number comes from the National Weight Control Registry, which reports that more than 87% of its members were still maintaining at least a 10% loss a decade later. That is a real finding from real data. It also cannot possibly mean what it is used to mean, for a reason that becomes obvious the moment you look at who is allowed to join.
This article is about what the registry does and does not show, which habits have been tested anywhere other than inside it, and the one finding about weight loss maintenance that survives every objection.
Quick fact: To enrol in the registry you must have already lost at least 30 pounds and already kept it off for at least a year. The 87% figure describes what happened next, to people selected for having already succeeded. It is not a success rate for anyone who starts a diet.
Two statistics, and what each one actually rests on
Both of the numbers that dominate the weight loss maintenance debate deserve the same scrutiny.
| The claim | Where it comes from | What it actually supports |
|---|---|---|
| “95% of people regain the weight” | Stunkard & McLaren-Hume, 1959 — 100 patients at one clinic. At two years, 2% had maintained a loss of 20 lb or more | That one clinic’s treatment, in 1959, mostly did not work. Nothing about the general population today |
| “About 20% succeed long-term” | A random digit dialling survey of 500 adults. Of 228 who were overweight or obese, 47 (20.6%) reported intentionally losing ≥10% and keeping it off a year or more | A single phone survey of self-reported weight history. Wing and Phelan, who cite it, say plainly that the data are limited and the definitions vary between studies |
| “87% of maintainers stay down at 10 years” | Nearly 3,000 registry members followed a decade — and the figure is an estimate, not a direct count | What happens to people who had already maintained a large loss for a year before enrolling |
None of these is a fabrication, and none of them answers the question people actually ask, which is what are my chances? The honest answer is that nobody has a trustworthy figure for that, because the studies that would produce it — following a large, representative group of people from the day they start — mostly stop at six or twelve months.
The survivorship problem at the centre of weight loss maintenance research
The National Weight Control Registry was founded by Rena Wing at Brown University and James Hill at the University of Colorado, and is coordinated through the Miriam Hospital in Providence. It takes adults who have lost at least 30 pounds and kept it off at least a year, then surveys them annually for as long as they stay enrolled. More than 10,000 people have joined. The average member arrived having lost around 66 to 70 pounds and having held it for five and a half years.
It was designed to answer a question trials rarely reach: not does weight loss work, but what do the people it worked for actually do? For that purpose it is genuinely valuable, and there is no substitute for it.
But the 87% figure cannot be used as evidence that maintenance is likely, and the reason is worth stating slowly, because almost every article that quotes it gets this wrong.
Everyone in the registry was admitted on the condition of having already succeeded. The people whose weight came back in month eight are not in the dataset. They were never eligible. Measuring how many registry members are still down at ten years and concluding that maintenance usually works is like measuring how many marathon finishers can run 10 km and concluding that most people can run a marathon.
There is a second layer on top of the first. A registry about successful maintenance is a registry people are more likely to leave quietly if they regain. The 87% is drawn from those still responding at follow-up, which filters the same way again.
The authors of the registry do not claim otherwise — this is a criticism of how the number travels, not of the people who produced it. What the registry genuinely demonstrates is narrower and still worth having: long-term weight loss maintenance is possible, it is not vanishingly rare, and the people who achieve it share identifiable habits. That is a menu, not a probability.
Who is actually in the registry
The word “self-selected” does less work than the actual figures, which the registry’s own authors publish.
| Characteristic | Figure |
|---|---|
| Women | 77% (other summaries put it at 80%) |
| College educated | 82% |
| Caucasian | 95% |
| How members were recruited | Mainly through newspaper and magazine articles |
| Weights | Self-reported, verified only in a subgroup; the authors note they slightly underestimate actual weight |
| Dietary intake reporting | Typically underestimated by 20–30%, by the authors’ own assessment |
| Members who say they are still trying to lose weight | 55% |
Read the last row twice. A majority of the people held up as having successfully finished losing weight describe themselves as still trying to lose weight. That is not a gotcha — it is a real and under-discussed feature of what maintenance looks like from the inside, and it sits awkwardly beside the framing of the registry as a group of people who arrived somewhere.
The 95% figure in the third row matters for a different reason: whatever these habits are doing, they have been observed almost entirely in white, college-educated women. That does not make the habits wrong. It means nobody should present them as established across populations they were barely measured in.
For a sense of who the typical member is: the average woman in the registry is about 45 years old and weighs around 145 pounds; the average man is about 49 and weighs around 190. The entry requirement is simply being an adult who has lost 30 pounds and held it a year. And at the two-year mark, 96% of participants remained at least 10% below their highest lifetime weight — a figure that sounds remarkable until you remember that holding a loss for one year was the condition of being counted at all. Every impressive number this registry produces carries that same qualification, which is why the research on weight loss maintenance reads so much more optimistically than the experience of most people who attempt it.
What members actually report — corrected
Here the earlier version of this article, and most coverage of the registry, overstated things. The real percentages are more interesting than the rounded ones.
| Behaviour | How it is usually reported | What the data says |
|---|---|---|
| Physical activity | “~2,600 kcal/week” | Women 2,545 kcal/week; men 3,293. The single number usually quoted is roughly the women’s figure. Both equate to about an hour a day of moderate activity, most commonly brisk walking |
| Self-weighing | “Regularly, most commonly once a week” | 44% weigh at least once a day; 31% at least once a week. Daily is the modal behaviour, not weekly — and this matters, for reasons below |
| Breakfast | “Nearly all members eat breakfast” | 78% eat breakfast every day; 4% never do. A clear majority, not “nearly all” |
| Weekday/weekend consistency | “Members keep the same pattern all seven days” | 59% report the same eating; 39% are stricter on weekdays. Nearly four in ten do run a tighter week than weekend |
| Holidays | Rarely mentioned | 45% eat the same; 52% are stricter outside holidays |
| Method of initial loss | Varied | 45% lost weight alone, 55% used a programme. 98% changed what they ate; 94% increased activity |
The consistency row is the one worth rescuing rather than discarding, because underneath the overstatement there is a real finding with an effect size attached: registry members with a consistent weekly eating pattern were 1.5 times more likely to stay within 5 pounds over the following year. That is a useful thing to know. “Everyone does it the same all seven days” was not.
Which weight loss maintenance habits have been tested outside the registry
This is the question that decides how much weight to put on each item, and it is the one the registry cannot answer about itself. The registry tells you what maintainers do. Only a trial can tell you whether doing it causes anything — and for most of the weight loss maintenance advice built on this dataset, no such trial exists.
| Habit | What the registry shows | What independent evidence shows |
|---|---|---|
| About an hour a day of activity, mostly walking | Strong, consistent association | Supported. A 2026 meta-analysis of 14 randomised trials and 3,758 participants found that people who raised their daily step count and sustained it regained less. Separately, in the CALERIE trial, taking half of an energy deficit from added activity instead of all from food eliminated the compensatory decline in daily movement |
| Self-weighing | 44% daily, 31% weekly; declines in self-weighing preceded regain | Weak and unflattering. In a randomised trial of 183 adults, instructing daily weighing produced no significant additional weight loss over three months, and weighing more often was not associated with losing more — about 20 g per extra day, with a confidence interval spanning zero. The authors concluded it is ineffective as a sole strategy. Note the mismatch: that trial tested weight loss, not maintenance, so the registry’s claim is not refuted — it is simply untested |
| Eating breakfast | 78% daily | Contradicted. A 2019 BMJ review of randomised trials found no difference in metabolic rate, found breakfast skippers slightly lighter on average, and found that people assigned to eat breakfast consumed about 260 more calories a day overall |
| Consistent weekday/weekend eating | 1.5× more likely to stay within 5 lb | Untested. A plausible association with no trial behind it |
| Mostly home-prepared food | Commonly reported | Untested in this context — and the registry’s own dietary data is underestimated by 20–30%, so what members report eating is not reliable |
| A larger initial weight loss | Associated with better 10-year outcomes | Untested as a strategy. It may simply be that people whose bodies and circumstances allowed a large loss are the people whose circumstances allowed maintenance |
The self-weighing row deserves a closer look, because its trial is weaker than a null result usually sounds. That study randomised 183 adults for three months, was not powered to detect the half-kilogram difference it actually observed, lost its objective scale data to a technical failure so that weighing frequency came partly from returned record cards — of which only 55% came back — and had a twelve-month follow-up planned and then abandoned once nothing appeared at three months. Its own confidence interval for the between-group difference is reported alongside a label of “not significant” in a way that does not obviously agree. So the fair summary is not “self-weighing does not work.” It is that one small, short, underpowered trial found nothing, that nobody has run the trial that would test the registry’s actual claim, and that the registry’s association is therefore unverified rather than refuted.
The breakfast row is the clearest demonstration of why this column exists. Breakfast is the registry’s most-quoted habit, and the trial evidence points the other way. That does not mean breakfast is bad for you — it means an association inside a self-selected group of people who were already succeeding is not a reason to start eating breakfast, and the earlier version of this article was right to flag the caveat and wrong to leave it at a caveat.
The one weight loss maintenance finding that holds up best
Strip away everything that depends on self-selection and one result is left standing, and it happens to be the most useful one for anybody in the difficult part.
Among registry members, the risk of regaining weight falls sharply with how long you have already maintained. Taking two years of maintenance as the reference point:
| Time already maintained | Odds of gaining 5 lb rather than maintaining |
|---|---|
| 2 years (reference) | 1.00 |
| 2 to 4.9 years | 0.42 |
| 5 years or more | 0.29 |
After two years, the authors put the reduction in regain risk at nearly 50%. This is still observational and still drawn from the same selected group — but it is a within-group comparison rather than a claim about the general population, which is exactly the kind of question the registry is built to answer. Comparing longer-tenured members to shorter-tenured ones does not depend on who was eligible to enrol.
The practical consequence is a genuine reframe. The first couple of years are not a preview of forever at constant difficulty. They appear to be the expensive part of a process that gets measurably cheaper. Both are hard; only one is finite.
What precedes losing ground
Two things, with unequal confidence behind them.
The first is well documented in the ten-year follow-up by Graham Thomas, Rena Wing and colleagues in the American Journal of Preventive Medicine: regain was preceded by declines in physical activity, declines in dietary restraint, declines in self-weighing, and increases in fat intake. Members who kept doing the things kept the weight off. Note that this is near-tautological as evidence of causation and still useful as a warning sign — the drift in behaviour showed up before the drift in weight.
The second is more specific and I can vouch for it less. Two papers nine years apart — Niemeier and colleagues in Obesity in 2007, and Lillis and colleagues in Obesity Science & Practice in 2016 — both report that internal disinhibition predicts weight regain: eating in response to internal cues such as emotions, thoughts or stress, rather than external ones such as the sight or smell of food. A finding replicated independently nine years later is worth more than a single result. I could not open either full text, so I am not quoting effect sizes, and I could not verify the common claim that external disinhibition showed no such effect.
If the direction is right, the implication is still worth sitting with: the durable risk is not the dessert tray at a party. It is the link between how you feel and whether you eat. That is a different problem from portion sizes, and it responds to different help. It is also, notably, the predictor that most closely resembles the thing the registry’s critics worry about.
The registry’s own limitations, in its authors’ words
The people who run the registry are more candid about its limits than most of the weight loss maintenance writing that cites it.
| Limitation | As stated |
|---|---|
| Self-selection | Members volunteer, recruited mainly via newspaper and magazine articles |
| Narrow sample | 77% women, 82% college educated, 95% Caucasian |
| Self-reported weight | Verified only in a subgroup; slightly underestimates actual weight |
| Self-reported diet | Intake typically underestimated by 20–30% |
| Observational design | Identifies associations; cannot establish that any habit causes maintenance |
| Mixed perspective | Survey data reflects a lifetime view and may fold together successful and unsuccessful attempts |
To that list belongs one more, which the registry’s critics raise rather than its authors. Some of the behaviours described here — daily weighing, high vigilance, sustained restraint, and 55% of members still actively trying to lose weight — are also features of disordered eating. Daily weighing is the single most common pattern in this group, not the weekly habit usually described, and presenting it as a neutral maintenance tool understates the concern. For many people a scale is information. For some it is not, and the difference is not something an article can assess from here.
When should you see a doctor?
Speak to a doctor or a registered dietitian if any of the strategies above have started to feel compulsive, if stepping on a scale produces real distress, if you find yourself avoiding the scale and then weighing repeatedly, or if you have any history of disordered eating — the vigilance that supports maintenance for some people becomes harmful for others, and that is a question for someone who can see your whole situation rather than for a list of habits. If weighing yourself causes distress, not weighing yourself is a legitimate choice, and the research in this article is nowhere near strong enough to override that.
Also worth raising: meaningful unexplained regain despite unchanged habits, which can involve medication effects, thyroid changes or other medical causes rather than behaviour.
Frequently asked questions
Is it true that 95% of people regain the weight they lose? No. That figure traces to a 1959 study of 100 patients at a single clinic, where 2% had maintained a 20 lb loss at two years. It has been repeated ever since and does not describe current evidence.
So what are my actual chances? Nobody has a figure you should trust. The most-cited optimistic estimate — about 20% — comes from one telephone survey of 228 overweight adults relying on self-reported weight history. The registry’s 87% describes people who had already succeeded before enrolling. The honest answer is that long-term weight loss maintenance is clearly achievable and nobody can give you a reliable probability.
How much activity do successful maintainers do? About an hour a day of moderate activity, most commonly brisk walking — reported as 2,545 kcal a week by women and 3,293 by men. That is considerably more than standard weight-loss advice suggests, and it is the registry’s best-corroborated habit. NEAT and how many steps a day you need both cover how to get there without structured exercise.
Should I weigh myself daily? Daily is what most registry members do, and declines in self-weighing preceded regain in the ten-year follow-up. But a randomised trial found that instructing daily weighing did not produce additional weight loss, and nobody has tested it properly as a maintenance tool. If it functions as information for you, there is a reasonable case for it. If it functions as a verdict, there is no evidence strong enough to justify the cost.
Does eating breakfast help keep weight off? 78% of registry members eat it daily, and randomised trials point the other way — no metabolic advantage, and people told to eat breakfast ate about 260 calories a day more. Eat breakfast if you want to. Do not take it up as a maintenance strategy on this evidence.
Does maintenance get easier? This is the registry’s strongest finding. Compared with two years of maintenance, the odds of regaining 5 lb were 0.42 at two to five years and 0.29 beyond five. The hard part appears to be finite.
Why do people regain after losing successfully? Behaviour drift comes first — activity, restraint and self-weighing all declined before weight returned. There is also a physiological side covered in breaking a weight loss plateau and weight cycling, including the unconscious decline in daily movement that accompanies a sustained deficit.
The bottom line
The registry is usually presented as evidence about the odds, and used either to reassure people or to tell them what to do. It is neither, and it is more useful once you stop asking it to be.
What it is, properly understood, is a description of a group — 10,000 mostly white, mostly college-educated, mostly female volunteers who did something difficult and then answered questionnaires about it for years. Read as a probability it is meaningless, because everyone in it was admitted for having already succeeded. Read as a prescription it overreaches, because its most-quoted habit is contradicted by trials and most of the others have never been tested at all.
Read as a description of what weight loss maintenance looks like for the people who manage it, it is worth a great deal. These people exist, in large numbers, which is more than the folklore allowed. They move roughly an hour a day. Their eating is consistent rather than cyclical. Most of them watch the scale, and more of them than is usually admitted are still fighting. And whatever they are doing gets measurably easier after about two years — which is the one thing here that is both well supported and genuinely worth knowing in advance.
This article is for general informational purposes and is not a substitute for personalised medical or nutritional advice. Consult a healthcare provider or a registered dietitian before adopting any of the strategies described, particularly if you have a history of disordered eating. If weighing yourself or monitoring food has become a source of distress, that deserves support in its own right, and nothing in this research is a reason to continue.
References
- Wing RR, Phelan S. Long-term weight loss maintenance. American Journal of Clinical Nutrition, 2005;82(suppl):222S–225S — the 1959 Stunkard & McLaren-Hume figures, the 20.6% telephone-survey estimate, the sex-split activity figures, the self-weighing, breakfast, weekday/weekend and holiday percentages, the 1.5× consistency association, the maintenance-duration odds ratios, and the registry’s stated limitations. Full text
- Thomas JG, Bond DS, Phelan S, Hill JO, Wing RR. Weight-loss maintenance for 10 years in the National Weight Control Registry. American Journal of Preventive Medicine, 2014;46(1):17 — nearly 3,000 members over a decade, the estimated 87% figure, and the behaviour declines that preceded regain. Summary
- Madigan CD, Jolly K, Lewis AL, Aveyard P, Daley AJ. A randomised controlled trial of the effectiveness of self-weighing as a weight loss intervention. International Journal of Behavioral Nutrition and Physical Activity, 2014;11:125 — 183 adults, three months, and the null result for instructed daily weighing
- Sievert K, et al. Effect of breakfast on weight and energy intake: systematic review and meta-analysis of randomised controlled trials. BMJ, 2019 — no metabolic-rate difference, breakfast skippers slightly lighter, and the ~260 kcal/day higher intake in breakfast groups
- Niemeier HM, et al. Internal disinhibition predicts weight regain following weight loss and weight loss maintenance. Obesity, 2007; and Lillis J, et al. Internal disinhibition predicts 5-year weight regain in the National Weight Control Registry. Obesity Science & Practice, 2016 — cited for the finding stated in their titles; neither full text was reachable, so no effect sizes are quoted here
- National Weight Control Registry overview — founding, enrolment criteria, membership numbers, average loss and maintenance duration, and the 45%/55% split between losing alone and using a programme
