Set Point Theory: Why Your Weight Moves, Why It Pushes Back, and Why That Isn't a Verdict on You

Two fears, one idea

Two moments on the scale frighten people more than any others. The first is small and happens most mornings: you did nothing different, and the number jumped. The second is bigger and slower. You've lost weight, maybe on a GLP-1 medicine, and you're scared of what happens when you stop.

Both fears lead back to the same question. Does my body have a weight it wants to be, and will it drag me back there?

That's what people mean by set point. You'll hear it used as comfort ("your body knows where it wants to be") and as doom ("your body will always win"). Neither is quite right. The science is more interesting than both, and once it's laid out, most of the fear has somewhere to go.

This post covers where the idea came from, the evidence that the body really does defend its weight, the serious scientific arguments against a simple set point, why the scale moves by amounts that aren't fat, how often it makes sense to weigh, and what the trials show happens when GLP-1 medicines stop. We've tried to be honest about what's settled and what isn't. Quite a lot isn't.

A quick note on names, because brands differ by country. Tirzepatide is sold as Mounjaro in the UK, Ireland and Australia, and as Zepbound in the US. Semaglutide for weight is Wegovy. We'll mostly use the generic names.

Where the idea came from

In 1953 a British physiologist, Gordon Kennedy, proposed that the brain keeps track of how much fat the body is storing and adjusts eating to keep it steady (Kennedy, 1953). It was an idea built on experiments in rats, and it turned out to be a good one.

Over the following decades, researchers found they could change the weight animals held by damaging specific spots in the hypothalamus, the part of the brain that looks after hunger, thirst and temperature. After damage to one area, rats settled at a lower weight and stayed there (Keesey and Hirvonen, 1997). That looked a lot like a thermostat with its dial turned down. The thermostat picture stuck, and "set point" became its name.

Then in 1994 a lab at Rockefeller University in New York found a messenger. A gene active in fat tissue makes a hormone, soon named leptin, that tells the brain how much energy is in storage. Mice with a broken copy of the gene become profoundly obese (Zhang et al., 1994). For a while it looked as though the whole system had been found.

Two things from this history are worth holding on to. Set point started as a model, a way of explaining what researchers saw. And the scientists who championed it said from the start that the point could shift over a lifetime (Keesey and Hirvonen, 1997). It was never meant to be a number fixed at birth.

The evidence that the body defends its weight

The best-known test was run in New York in the early 1990s. Rudolph Leibel, Michael Rosenbaum and Jules Hirsch took 41 people, 18 with obesity and 23 who had never had obesity, and measured exactly how much energy each of them used at their usual weight. Then they changed everyone's weight, down by 10 to 20% or up by 10%, held it there, and measured again (Leibel, Rosenbaum and Hirsch, 1995).

Held at a lower weight, people used less energy than their smaller bodies should have needed. Held at a higher weight, they used more. Both groups responded the same way. The authors' own summary is careful and worth quoting: keeping weight lower or higher "is associated with compensatory changes in energy expenditure, which oppose the maintenance of a body weight that is different from the usual weight."

A decade later the same group tried something clever. They held ten people at a reduced weight, then gave them small doses of leptin to bring it back up to where it had been before. Energy use, how efficiently their muscles worked and their thyroid hormones all went back to pre-weight-loss levels (Rosenbaum et al., 2005). The body had been reading the fall in leptin as a shortage and responding to it.

Appetite pushes back too, and for longer than most people expect. In Melbourne, Priya Sumithran and colleagues put 50 people on a ten-week very-low-energy diet and then followed them. A year after the weight loss, the hormone that drives hunger (ghrelin) was still higher, and several that help you feel full (leptin, peptide YY and cholecystokinin among them) were still lower, than before the diet. People said they felt hungrier, too (Sumithran et al., 2011).

How strong is that pull? One analysis used a neat trick. A diabetes medicine called canagliflozin makes the body lose some energy in urine without the person noticing, so any change in how much people then eat is their body's response rather than a decision. Working backwards from their weight, researchers estimated that as weight came off, people ended up eating noticeably more, by an amount more than three times the size of the drop in energy used (Polidori et al., 2016).

If we could make one point stick from this whole post, it's this one. After weight loss, being hungrier is the expected outcome. It has been measured again and again. It is not a sign that you've lost your discipline. (Our piece on why weight loss slows and stalls on a GLP-1 picks up the same thread.)

Then there's the study everyone has heard of. Researchers followed 14 contestants from the US TV show The Biggest Loser for six years. Their resting metabolism was still well below where it had started, by more than their change in size explained (Fothergill et al., 2016). The headline wrote itself: your metabolism is broken for good.

The fine print tells a different story. These were people who lost a very large amount of weight very fast, under extreme conditions. In the same study, the people with the biggest slowdown were the ones who had kept the most weight off, and the authors describe the slowdown as a "proportional, but incomplete, response" to ongoing efforts to hold weight down. A follow-up found that the contestants who kept weight off had greatly increased their everyday physical activity (Kerns et al., 2017). That's a headwind. It isn't a wall.

And some of the slowdown may be temporary. In a 2020 study of 71 people who lost weight on a low-calorie diet, the metabolic slowdown was about halved once their weight had been stable for a month, and it didn't predict who regained weight a year later (Martins et al., 2020). The paper's title calls metabolic adaptation "an illusion", which other researchers disputed in the same journal. The fair reading is that part of what looks like a stubborn metabolism is a body that's still in the middle of losing.

The push works the other way too, though less predictably. When volunteers were overfed by the same large amount for eight weeks, the fat they stored varied tenfold from person to person, and most of the difference came from how much extra they moved without meaning to: fidgeting, standing, getting up more (Levine, Eberhardt and Jensen, 1999). When 12 pairs of identical twins were overfed by the same amount for about three months, the biggest gainer put on around three times as much as the smallest, and each twin responded much more like their own twin than like anyone else (Bouchard et al., 1990).

That twin result points to genes. Across 88 twin studies, between roughly half and nine-tenths of the differences in body size between people came down to genetic differences (Elks et al., 2012). In a Danish study of adults who had been adopted as children, body size tracked their biological parents and not the parents who raised them (Stunkard et al., 1986). Genes shape where your weight tends to settle. As the next section shows, they don't set it in stone.

Where set point theory runs into trouble

If the body defended one fixed weight, the average weight of a whole population shouldn't change much from one generation to the next. It has. Since the 1980s, body weight has risen across many countries, far too quickly for genes to have changed. A group of 25 researchers put it plainly in 2011: the set point model "does not effectively explain the 'obesity epidemic'" (Speakman et al., 2011).

There's also a simpler explanation for why weight looks defended, and it's older than you'd think. In 1977 two psychologists showed that a basic feedback system with no set point at all could explain the evidence that had been used to argue for one (Wirtshafter and Davis, 1977). Think of a lake rather than a thermostat. The water level stays fairly steady because more flows out when it's high and less when it's low. Change the rainfall and the lake settles at a new level. Nothing is defending a particular number. This became known as the settling point model: your weight settles wherever what you eat and what your body uses come into balance, given your genes and your surroundings.

Neither model fits everything. Set point struggles with the obesity epidemic. Settling point struggles with the biology: the hunger hormones that stay shifted for a year, the leptin that switches energy use back on (Speakman et al., 2011).

Newer ideas try to bridge the gap. One observation that comes up again and again is that the body seems to resist weight loss more strongly than weight gain (Müller, Bosy-Westphal and Heymsfield, 2010), which fits most people's experience. The dual intervention point model, from John Speakman, says there isn't a single point but a range. Inside it, your weight drifts with your food, your routine and your life, and the body barely reacts. Move beyond the edges of the range and physiological defences switch on (Speakman and Hall, 2023).

The researchers don't agree on which picture is closest. Kevin Hall, at the US National Institutes of Health, compared the models in 2017 and concluded that the set-point model was "most commensurate with current data" (Hall and Guo, 2017). A 2018 review took the opposite emphasis: "there is no consensus framework" for how body weight is controlled (Müller et al., 2018). And in 2023 Speakman and Hall, who have championed different models, wrote together that "further experiments to test between the models are sorely required" (Speakman and Hall, 2023).

We find that reassuring, oddly. When the people who know this best are still arguing, anyone who tells you exactly what your body will do is guessing.

So is there a set point?

Here's where we've landed. It's less tidy than either the comforting version or the frightening one.

The body does push back when weight moves, especially downwards. That part has been measured in people, repeatedly. It shows up as hunger that stays higher, as a body that uses a little less energy, and as food taking up more room in your head. All of that is physiology. None of it is a flaw in you.

What's still argued about is whether there's a fixed point. The evidence fits better with a range your body is fairly relaxed about, which gets nudged over the years by your genes, your age, your surroundings and some medicines. "Set point" is a handy shorthand for that range. It isn't a number written into you.

That matters in both directions. "Your body will always drag you back to your heaviest weight" goes further than the evidence does. So does "just push through it with willpower", which ignores something real and measurable. Both ideas tend to leave people feeling worse about themselves.

One part of the set point story is on very solid ground, and it's the part eating disorder services care about most. When someone eats less than their body needs for a long time, food takes over their thoughts. In the Minnesota starvation experiment during the Second World War, 36 healthy men volunteered for a long period of semi-starvation. Many became apathetic, irritable and exhausted (Kalm and Semba, 2005), and reviews of that work and later dieting research describe the same pattern: preoccupation with food, low mood, poor concentration, and eating binges once food is freely available again (Polivy, 1996). They were healthy when they started. The preoccupation came from the restriction. If food thoughts are crowding out the rest of your life, that's worth knowing, and it's worth talking to someone about. Our piece on how restriction and bingeing feed each other goes further.

Why the scale moves when your fat doesn't

Now for the daily fear. Body fat changes slowly. The number on the scale changes quickly, and most of what it shows from one day to the next is water, stored carbohydrate and whatever is still passing through you.

How big is the wobble? One of the longest records we have comes from one healthy man who weighed himself every morning, under the same conditions, for almost 30 years: 9,521 days of data. From one day to the next his weight changed with a standard deviation of about half a percent of his body weight, and that stayed true across all three decades (Schneditz et al., 2023). The researchers put it down mainly to water. Half a percent sounds tiny. For most adults it's somewhere around a pound (about half a kilo), and every so often a change of twice that is still perfectly normal.

Here's what's usually behind it.

Stored carbohydrate is the big one. Your muscles and liver keep carbohydrate as glycogen, and each gram is stored with at least three grams of water (Fernández-Elías et al., 2015). After a stretch of eating less, the body can pack away roughly half a kilo (about a pound) more glycogen before extra carbohydrate starts being turned into fat (Acheson et al., 1988). Put those two findings together and a few carb-heavy days can add more than a kilo (over 2 lb) of glycogen and water without any fat at all. It works in reverse too, which is why the first week of a low-carb diet looks so dramatic, and why that weight comes back as soon as the bread does (Kreitzman, Coxon and Szaz, 1992).

Then there's everything still inside you. A litre of water weighs a kilo (2.2 lb) until your body has dealt with it, and food is the same. Pooling studies from around the world, the typical day's stool weighs around 128 g (about 4.5 oz) (Rose et al., 2015), and the time food takes to pass through can vary a lot even in the same healthy person tested twice (Degen and Phillips, 1996). A bigger meal the night before, or a slower day for your gut, shows up in the morning.

Salt is the explanation people reach for most, and the real picture is messier than the usual advice. In a tightly controlled study where men ate very different amounts of salt, total body water and body weight didn't go up, although fluid moved around inside the body (Heer et al., 2000). The 30-year daily record, on the other hand, links much of the daily wobble to water that tracks sodium (Schneditz et al., 2023). So salt probably plays a part alongside everything else, and we'd be wary of anyone who tells you exactly how much a takeaway adds.

Your menstrual cycle matters too, if you have one. In a study that measured 42 women twice a week across a cycle, weight was about 0.45 kg (1 lb) higher during a period than in the first week of the cycle, and that difference was extra fluid, not fat (Kanellakis et al., 2023). In a year-long diary study, the feeling of fluid retention peaked on the first day of bleeding (White et al., 2011).

And there's the week itself. In 1,421 adults across the UK, Denmark and Portugal who weighed themselves on smart scales, weight was highest on Mondays, fell through the week to its lowest on Fridays, then climbed again over the weekend. The swing was about 0.35% of body weight (Turicchi et al., 2020). An earlier study of 80 adults found the same pattern, and its authors said it plainly: weekend-to-weekday variation "should be considered as normal instead of signs of weight gain" (Orsama et al., 2014). The same European study found Christmas added about 1.35% on average, and not all of it had gone by March, which is a gentler, slower pattern than the "it'll drop straight off" you sometimes hear.

Put all of that together and you can see why one morning's number isn't worth much on its own. It's a real measurement. It just isn't a measurement of fat.

Daily, weekly, or not at all?

This is where the advice seems to contradict itself, so it's worth slowing down.

In weight-management research, frequent weighing looks helpful. In the STOP Regain trial, 314 people who had recently lost weight were taught a programme built around weighing every day. The face-to-face group regained less over 18 months than a comparison group who were sent newsletters (Wing et al., 2006). Another trial gave people smart scales, an online weight graph and weekly feedback emails, and they lost more over six months than a group who started later (Steinberg et al., 2013). Both teams looked specifically for psychological harm, such as low mood, binge eating and body dissatisfaction, and didn't find it in their participants (Wing et al., 2007); (Steinberg et al., 2014). A systematic review came to the same conclusion (Zheng et al., 2015). Two details from a meta-analysis of the trials are easy to miss, though. Weighing on its own, without the rest of a programme, hadn't been shown to work. And trials that told people to weigh daily did no better than trials that said weekly (Madigan et al., 2015).

In eating disorder treatment the advice is different. Checking your weight again and again is recognised as one of the ways an over-focus on weight and shape shows itself, and so is avoiding the scale altogether (Shafran et al., 2004). Cognitive behavioural therapy, the talking therapy with the strongest evidence for eating disorders, handles this by weighing people once a week, in the session, with the therapist. Weighing in between is discouraged, and the number is always read across several weeks rather than one reading at a time (Waller and Mountford, 2015), describing the approach in Fairburn's CBT-E manual. The Centre for Clinical Interventions in Western Australia explains the thinking in its patient handouts on weekly weighing and on set point, which were the starting point for this piece: weighing constantly feeds the preoccupation, while never weighing leaves the fear untested and growing (CCI). There's more on how CBT for eating disorders works in our guide.

So who's right? Both are, for their own group. The weight-loss trials recruited adults who wanted to lose weight, measured harm across the group as a whole, and paired the number with a trend graph and support. They weren't designed to show what frequent weighing does to someone whose relationship with food is already struggling. In eating disorder care the goal is close to the opposite: to loosen the grip a number has on your mood, your food and your sense of who you are. And there are signs that frequent weighing isn't harmless for everyone. In two long-running US studies of young people, frequent self-weighing in teenage girls and young women went along with more weight concern, lower self-esteem and, in one of them, more binge eating and unhealthy weight-control behaviour years later (Neumark-Sztainer et al., 2006); (Pacanowski et al., 2015). The meta-analysis that found no harm on average also found that the effect depends on who is doing the weighing (Benn et al., 2016).

A rule of thumb, then, rather than a rule. If a number on the scale changes your mood, your plans for the day or what you let yourself eat, weighing once a week, on the same morning under the same conditions, and reading it as a trend over a month, is the gentler option. Britain's NHS weight-loss plan says once a week too (NHS, United Kingdom), and it is the only national plan we have checked, so take it as one country's advice rather than a rule everywhere. If you're being treated for an eating disorder, your team's plan comes first, whatever any article says, this one included.

It's also why Healthcount keeps your daily weight off its home screen and shows it on the Trends page instead, as a line across weeks rather than a single number over breakfast.

Coming off a GLP-1: the body's defence, in trial data

This is where the science stops being abstract.

GLP-1 medicines turn appetite down. In one trial, people on semaglutide ate about 35% less at a lunch where they could have as much as they liked, and reported fewer and weaker cravings (Friedrichsen et al., 2021). Kevin Hall's modelling of the big trials suggests these medicines do something a diet on its own doesn't: while you're taking them, they weaken the appetite feedback described earlier, the pull that makes you hungrier as weight falls, by roughly 40 to 70%. That's why weight keeps falling for longer on the medicine before it levels off (Hall, 2024). He describes the results as preliminary, and they come from a model rather than a direct measurement.

In set point language, the medicine moves the level your weight settles at while you're taking it. Stop it and the old pull comes back. The withdrawal trials show that clearly.

At the end of the STEP 1 trial, semaglutide and the trial's diet and activity support stopped together, and 327 people were followed for another year. The group who'd had semaglutide had lost an average of 17.3% of their starting weight. A year after stopping, they'd regained 11.6 percentage points of it, about two-thirds, and were on average 5.6% below where they'd begun. Blood pressure, blood sugar and other markers drifted back towards where they started, too (Wilding et al., 2022).

In STEP 4, 803 people who had lost an average of 10.6% over 20 weeks on semaglutide were split into two groups without being told which was which. One carried on; the other switched to a placebo. Both kept the same lifestyle support. Over the next 48 weeks the group who carried on lost a further 7.9%, while the placebo group's weight went up by 6.9% (Rubino et al., 2021).

SURMOUNT-4 did the same with tirzepatide. After 36 weeks on the medicine, 670 people had lost an average of 20.9%. Over the following year, weight rose by 14.0% in those switched to placebo and fell by a further 5.5% in those who carried on. Measured from the very beginning, the placebo group finished 9.9% lighter than they started, and the group who carried on finished 25.3% lighter (Aronne et al., 2024).

A 2026 review in the BMJ brought together 37 studies of people stopping weight-loss medicines. After semaglutide or tirzepatide, weight came back at an average of about 0.8 kg (1.8 lb) a month, and the authors projected a return to starting weight roughly a year and a half after stopping (West et al., 2026). That projection runs past the data, which the authors say themselves: the studies of those two medicines followed people for a year at most, and the model assumed weight would keep returning in a straight line.

What this does mean, and what it doesn't

It does mean that weight coming back after stopping is the usual outcome, on average, and that it's biology. The appetite that returns is the same feedback the medicine was quieting. That's not a character flaw.

It doesn't mean everyone goes back to where they started. In the STEP 1 extension, 95 of the 197 people who'd had semaglutide and came to the final visit were still at least 5% below their starting weight a year after stopping (Wilding et al., 2022). In SURMOUNT-4, 54 of the 308 people switched to placebo who had lost at least 10% regained less than a quarter of it, and in that group waist size, non-HDL cholesterol and fasting insulin hadn't changed significantly a year on, although blood pressure had still gone up (Horn et al., 2026). An average hides a wide spread.

It doesn't mean the weight comes back overnight. In every one of these trials it built up over months. A slow change is one you can see coming, if you're looking at a trend rather than one morning.

It doesn't tell us about other ways of stopping. These trials stopped the medicine completely at a fixed point, because they were designed to measure what the drug itself does. How and when to stop is a decision for you and your prescriber, and this post won't suggest a plan for it.

And, honestly, it doesn't show that diet and exercise can hold the line on their own. In STEP 4 and SURMOUNT-4 the lifestyle support carried on in both groups, and weight still came back on placebo (Rubino et al., 2021); (Aronne et al., 2024). The BMJ review found no evidence that the kind of support offered after stopping changed how quickly weight returned, although only a few of its studies compared different kinds of support (West et al., 2026). We'd rather tell you that than promise something the evidence doesn't back. What helps is knowing what's likely, watching the trend calmly, and talking your plans through with your prescriber before you need to.

If you're working through that decision, coming off a GLP-1 without regaining and quitting GLP-1s: what actually happens go further into the practical side.

A calmer way to read the number

If you weigh yourself, a few habits take a lot of the fright out of it.

• Compare like with like: the same scale, the same time of day, the same state (after the loo, before breakfast).
• Read the trend across four weeks, not the reading from one morning.
• Expect the weekly wave. Heavier on Monday than on Friday is normal.
• Expect your cycle, if you have one. The rise around a period is mostly fluid.
• After a salty, carb-heavy or late meal, give it several days before reading anything into the number.
• If a reading changes what you let yourself eat that day, take that as a sign to weigh less often, not to eat less.
• If you're coming off a GLP-1, keep the trend going, and agree with your prescriber in advance what change would be worth a conversation.

And remember what the scale can't measure: how you're sleeping, how strong you feel, whether food is taking up less room in your head than it used to. Those count too.

If one number on one morning has ever wrecked your day, a kilo up after two holidays and your period, your weight and your mood walk through two of the most common causes in more detail.

If this has stirred something up

Reading about weight can bring things to the surface. If it has, you don't have to sort it out alone.

Your doctor is a good place to start too. A helpline is not an emergency service: if you're in danger right now, call your local emergency number.

The questions people ask

This is information, not medical advice. Decisions about starting, continuing or stopping a prescribed medicine belong with the prescriber who knows your history. If you're receiving eating disorder treatment, follow your care team's guidance on weighing.