why am i not losing weight if i'm eating less.

Published September 14, 2026·Nutrition · Deficit science·17 min read

Eating less and not losing weight almost never means a broken metabolism. In the measured cases it means one of four boring things: the deficit is smaller than the log says, the exercise is worth less than the watch says, water is hiding the fat you did lose, or the deficit only exists Monday to Thursday. Here are the numbers on each, and the fourteen-day audit that tells you which one is yours.

let me guess. you have been "good" for five weeks. you eat less than everyone you know. your lunch is a salad, your snacks are protein bars, you say no to the office cake, you did the class, you hit the steps. and this morning the scale said the same thing it said on the first of the month, so you stood on it again with your feet in a slightly different place, and then you typed it into your phone at one in the morning and every page told you your metabolism had "adapted" and your body was "holding on" and your hormones were "out of balance", and all of that felt true, because it explained the thing without it being your fault. none of those pages cited a trial. i went and read the trials. twenty-eight of them, listed at the bottom.

i am not neutral about this one. i spent years at 1,200 calories, training hard, calling every stall a discipline problem and answering it by eating less. so i want to be careful about how i say the next part, because it is going to sound like the internet's least favourite answer. the trials say the deficit is usually not there, or not there every day, or it is there and the scale cannot see it yet. and "usually" has a number attached, which is the point of this file. the short version is in the box. the four reasons come after it, with the boring fixes, and the audit that tells you which one you have is at the end.

reason one: the deficit is on paper.

start with the study that should be on page one and is not, because it is from 1992 and it is rude. Lichtman and colleagues took people presenting for obesity treatment who said they had been eating under 1,200 calories a day and could not lose, which is the exact sentence this file is about, and put ten of them, nine women and one man, through fourteen days of indirect calorimetry and body-composition measurement. indirect calorimetry means measuring the oxygen you breathe in and the carbon dioxide you breathe out to calculate what you actually burn. their total energy expenditure and resting metabolic rate were within 5 percent of what their body composition predicted. nothing was slow. what they found instead was that the group under-reported what they ate by an average of 47 percent and over-reported their physical activity by 51 percent. the authors' conclusion is one sentence and it has aged perfectly: the failure to lose "is due to an energy intake substantially higher than reported and an overestimation of physical activity, not to an abnormality in thermogenesis."

i want to be fair to the ten people, because the internet has spent thirty years being unfair to them, and because i was one of them in every way except the lab. they did not lie. they described their eating as "relatively normal" and believed it. under-reporting is not a character flaw, it is what humans do with food when nobody is weighing it, and the size of the gap scales with how much you want the number to be low. that is the whole reason nutrition science stopped trusting food diaries and started using a method called doubly labelled water, where you drink water tagged with harmless heavy isotopes and the rate they leave your body tells researchers your real energy expenditure to within a few percent. when Trabulsi and Schoeller reviewed every self-report instrument against it, none of them came out accurate, and the bias was systematic, not random: it pointed downward, and it pointed downward more in the people who most wanted to lose weight.

the "including dietitians" is not a joke. Champagne 2002 took ten registered dietitians and ten women of the same age and weight, trained all of them to keep a seven-day weighed food record, told them it would be checked against doubly labelled water, and checked it. the dietitians still under-reported by 223 calories a day. the non-dietitians under-reported by 429. everyone knew they were being measured. everyone was weighing their food. the professionals were off by the size of a snack and the rest of us by the size of a meal. so when i say "log honestly", i do not mean try harder. i mean assume you are 400 calories light on a good day and build the plan around that number instead of around the one in the app.

then there is the food itself, which does not hold still either. Urban 2010 bought 29 "reduced-calorie" items from quick-serve and sit-down restaurants and ten frozen supermarket meals and burned them in a lab. the restaurant items averaged 18 percent more calories than stated, the frozen meals 8 percent, some individual items ran to 200 percent, and once the free side dishes were counted the entrées delivered 245 percent of their stated number. the app, to its credit, is not the weak link. when five popular tracking apps were checked against a dietitian's reference software on the same twenty weighed food records, their calorie totals were not significantly different. the database is fine. the error is the human between the plate and the phone: the oil that was "a bit", the sauce that was not logged, the bite of someone else's, the restaurant plate entered as its menu description, the "light" label that was 18 percent heavy. none of this requires a broken metabolism. it just requires a normal one and a normal memory.

reason two: you are burning less than the watch says.

the second half of the Lichtman gap was exercise, and the industry has since built a device for over-reporting it automatically. Stanford tested seven wrist-worn trackers, Apple Watch and Fitbit among them, on 60 people, 31 of them women, against indirect calorimetry while they sat, walked, ran and cycled. six of the seven measured heart rate to within 5 percent while cycling. not one of them got calories burned within 20 percent, and the errors were larger for walking, for heavier bodies and for darker skin. the authors' phrase was "caution in the use of energy expenditure measurements as part of health improvement programs", which is a polite way of saying do not eat that number back. a 400-calorie workout on the wrist is somewhere between 320 and 480 in reality, and you do not know which end, and the direction that hurts is the one your brain will pick.

the deeper problem is that even the real number does not all count. Careau 2021 used the largest dataset of measured human energy expenditure that exists, 1,754 adults living ordinary lives, and asked how much of the energy people burn in activity shows up in their daily total. the answer was that the body compensates about 28 percent of it, by spending less elsewhere, so only about 72 percent of the extra calories from moving translate into extra calories burned that day. and the compensation was larger in people carrying more body fat, which the authors flag as a chicken-and-egg question they could not settle. Pontzer 2016 found the same shape in 332 adults across five countries: total daily expenditure rises with activity at the low end and then flattens at the high end, as the body pulls the total back toward a range it prefers. so the treadmill's 400 was maybe 320 by the watch, and maybe 230 by the end of the day. that is not nothing. it is about a third of what you thought.

where does it go? mostly into the fidgeting you stop doing. in 1999 Levine overfed 16 people by 1,000 calories a day for eight weeks and found that two-thirds of the increase in their daily expenditure was non-exercise activity thermogenesis, or NEAT, the energy of posture, pacing, fidgeting and the thousand small movements of a day, and that differences in NEAT explained a ten-fold difference in how much fat people stored. run that in reverse and you have a deficit: Leibel 1995 measured 41 people holding a weight 10 percent below their usual and found total expenditure down 6 to 8 calories per kilogram of fat-free mass per day, split roughly evenly between resting metabolism and non-resting movement. for someone carrying 45 kilograms of fat-free mass that is roughly 270 to 360 calories a day, and about half of it is that you sit stiller, take the lift, and pace less, without noticing. this is the honest version of "your metabolism adapted". it is real, it is measurable, and it is mostly your body quietly cancelling the walk you did not know you were taking.

so is "starvation mode" real or not.

here is where i have to be straight with you, because both camps on the internet are wrong in opposite directions. the slowdown is real. it is also small, for anyone not on a television show. the measurements, in order of how much weight they should carry with you:

Martins 2020 put 71 people with obesity on 1,000 calories a day for eight weeks. they lost about 14 kilograms. metabolic adaptation, meaning a resting metabolic rate lower than their new body size predicted, was 92 calories a day at week nine. after four weeks of eating at their new maintenance it had halved. at one year it was 7 calories a day and not significant. and it did not predict who regained. the same group's 2022 study followed 65 premenopausal women on 800 calories a day until they reached a normal BMI: a 16 percent loss, about 12.5 kilograms, over 155 days, with an adaptation of 46 calories a day. that 46 calories was a real predictor of how long it took them to get there, so it is not nothing. it is also the size of half an apple. one more honest line from that paper: measured adherence to the 800-calorie prescription averaged 64 percent. even in a study, with the food prescribed, the deficit on paper was not the deficit in the body.

Müller 2015 did the tightest version, in 32 men, three weeks at half their energy needs with whole-body MRI and every hormone measured. resting expenditure fell 266 calories a day, but only 72 calories of that was true adaptation once the lost tissue was accounted for, and it appeared within the first three days, tracking the fall in insulin and in body water rather than leptin or thyroid hormone. that is the mechanism: the early slowdown is mostly your body shrinking and drying out, not your engine changing. the number everyone quotes, the one that made "metabolic damage" a genre, is the Biggest Loser follow-up: 14 contestants, six years after losing an average of 58 kilograms in 30 weeks, still burning about 500 calories a day less than their bodies predicted. that is a real finding about the most extreme intervention ever televised. it is not a finding about you eating 1,500 calories for six weeks. Rosenbaum 2008 found that in people holding a 10 percent loss the drop does persist past a year, so the adaptation is not a myth. it is a headwind of about 50 to 100 calories a day for a normal diet, more for a huge loss, and a headwind slows you down. it does not blow you backwards. if the scale is flat, the wind is not why.

one more thing "eating too little" does do, reliably and immediately, and it is the reason i wish someone had shouted this at me in the 1,200 years. it costs you muscle, and muscle is the thing you were training for. i will show you the trial in a minute, because it is also the trial that explains why the scale cannot see what you are doing. and the rule of thumb that made you expect faster loss in the first place is wrong in a specific way: Hall 2008 showed that the "3,500 calories per pound" rule roughly holds for people starting with more than 30 kilograms of body fat and overstates the deficit needed for leaner people, because a bigger share of what they lose is lean tissue, which is cheap to lose and easy to lose. it is also why, for the same deficit, women typically lose fewer scale-kilograms than men of the same weight, since women carry more body fat and fat is dense in energy. read that as: the leaner you get, the more of a careless deficit comes out of the muscle, and the less the scale rewards you for it. the fix is protein and lifting, not a smaller number.

reason three: you are losing fat and the scale cannot see it.

now the reason that is not your fault at all, and the one i think is most often true for someone who lifts. the scale weighs everything: fat, muscle, the water in both, the food in your gut, the glycogen in your liver and legs, and the fluid your kidneys decided to keep this morning. only one of those is what you are trying to lose, and it is the quietest one. Heymsfield's review of the early phase of weight loss puts a shape on this: phase one has a half-life of under a week, runs four to six weeks, and is dominated by rapid changes in body water, glycogen and expenditure. that is the drop everyone gets in the first fortnight and the reason the third and fourth weeks feel like hitting a wall. you did not hit a wall. you finished the fast, mostly-water phase and entered the slow, mostly-fat one, which looks like nothing on a scale that swings a kilogram a day for other reasons.

the biggest of those reasons is glycogen, the carbohydrate your muscles and liver store for fuel. Kreitzman 1992 is the reference: glycogen is stored with three to four parts water, so a few hundred grams of it carries over a kilogram of fluid with it. cut carbohydrate and the scale drops fast and flatteringly. eat a big pasta dinner after a hard leg day and it comes back overnight, and you did not gain fat, you refilled a tank. the overnight file has the full arithmetic on this. add a period: in 765 menstrual cycles in 62 healthy women, self-rated fluid retention was lowest in the mid-follicular phase, rose across the eleven days around ovulation and peaked on the first day of flow, and it did not track oestrogen or progesterone levels, so you cannot predict it from a symptom, only from the calendar. [the period file is the long version.] add a new program: the first few sessions of unfamiliar lifting make muscles swell with fluid for days, which is why the first month of a new program can add scale weight while your jeans loosen. and add the dieting itself: in 121 women randomised for three weeks, restricting to 1,200 calories a day raised total cortisol output, and simply tracking calories raised perceived stress. cortisol is a hormone involved in how the body handles sodium and water, so a link from a hard diet to holding fluid is plausible physiology. i am flagging honestly that this trial measured the cortisol, not the water, so treat the last step as mechanism, not a measured result.

and then the part that matters most if you train. this is the trial i promised. Nedeltcheva 2010 put ten overweight adults, three women and seven men, through the same fourteen-day calorie restriction twice, once with 8.5 hours in bed and once with 5.5. with enough sleep they lost 1.4 kilograms of fat and 1.5 of fat-free mass. with short sleep they lost 0.6 kilograms of fat and 2.4 of fat-free mass. total weight lost: about 2.9 kilograms versus 3.0. the scale could not tell the two bodies apart. one of them had lost more than twice the fat, and the other had lost most of its weight as muscle, water and organ tissue, was hungrier, and burned less fat at rest. so when i say the scale is the wrong instrument, i do not mean it as comfort. i mean it literally reads the same number for a good outcome and a bad one. the chart is below.

the trial that turns this into a plan is Longland 2016. forty young men, a 40 percent deficit, four weeks, lifting plus intervals six days a week, and the only difference was protein at 1.2 or 2.4 grams per kilogram of bodyweight. the high-protein group gained 1.2 kilograms of lean mass and lost 4.8 of fat. the scale moved about 3.6 kilograms. the low-protein group gained 0.1 and lost 3.5, and their scale moved 3.4. two hundred grams of difference on the scale, and underneath it 1.3 kilograms more fat gone and a kilogram of new muscle. this was men and four weeks, so read it as the direction, not the dose. but it is the direction: if you lift and eat enough protein, the scale will under-report your progress for exactly the reason your food diary over-reports your virtue, because it is measuring the wrong thing. the waist measurement and the log do not have this problem. [the recomposition file and the protein file take this from here.]

reason four: the deficit exists monday to thursday.

the last reason is a calendar. Racette 2008 weighed 48 adults, 30 of them women, every morning for a total of more than 600 person-weeks while they either cut calories or exercised for a year, with food diaries and accelerometers both validated against doubly labelled water. at baseline people gained 0.06 kilograms per day on weekend days and nothing on weekdays, from eating more on saturdays and moving less on sundays. during the diet, they were in a real deficit monday to friday and stopped losing on weekends entirely, and the exercise group actually gained on weekends, 0.08 kilograms a day, because they ate more. the authors offer this as an explanation for "the slower-than-expected rate of weight loss" in weight-loss studies generally. it is also an explanation for yours.

the rhythm is visible in anyone who weighs daily. Orsama 2014 analysed 4,657 weigh-ins from 80 adults over up to 330 days and found a weekly wave: weight rises from saturday, peaks on sunday and monday, and falls from tuesday. the people who lost or maintained weight were the ones with the biggest weekday compensation, the ones who slowly gained were the ones whose tuesday-to-friday did not undo the weekend. the authors' conclusion is worth taping to the scale: "weight variations between weekends and weekdays should be considered as normal instead of signs of weight gain." the monday number is water, glycogen and a real weekend. it is not information. the friday-to-friday trend is.

here is the arithmetic that makes this the most common reason of all. a 500-calorie daily deficit monday to thursday is 2,000 calories. two weekend days at plus 1,000 each, which is one dinner out with the 18 percent label gap and the sides, one brunch, two drinks and a "few" fries, is 2,000 calories back. the weekly deficit is now zero and every single day of it felt like dieting. and it hides itself: the people who did this in Racette's trial were not weekend bingers, they ate a normal amount on a day with no structure. sleep does the same thing more quietly. a meta-analysis of 11 sleep-restriction trials found short sleep added 385 calories a day of intake with no change in what people burned, more of it from fat and less from protein. so a late friday and saturday can cost you the deficit on sunday and monday without a single decision being made. [the sleep file has the training side of that number.] and none of it is a metabolism story: when 23 adults were measured through the winter holidays with doubly labelled water, their expenditure did not change at all. they gained because they ate out more and felt less full after the same meal. the expenditure side is boring. it is the intake side that does everything.

but my hunger is out of control. isn't that hormonal.

yes, and it is not the reason you are not losing. it is the reason the deficit is hard to hold, which brings you straight back to reasons one and four. Sumithran 2011 followed 50 people through a ten-week very-low-energy diet and a 13.5-kilogram loss, then measured their appetite hormones a year later. at 62 weeks, ghrelin, the hunger signal, was still up, and leptin, peptide YY, cholecystokinin and insulin, the fullness signals, were still down, and rated hunger was still higher than before the diet. the body defends a lost weight by making you want food, not by refusing to burn it. that is the whole difference between the two camps: "hormones" do not stop fat loss in a deficit, they make the deficit smaller than you intended, in ways that do not feel like eating more. a bigger portion. a second coffee with milk. the bite you did not log. the saturday.

and this is where the lifting matters twice. once because the muscle you keep is the tissue you are trying to look like. and once because [the hunger file has the trials] the post-workout hunger that erases a session is a real, measurable pattern in about half of people, and it is not a matter of willpower but of what you planned to eat afterwards. a deficit that lives only in the calorie target is a deficit that gets negotiated three times a day. a deficit that lives in a weighed plate and a planned post-session meal does not have to be re-decided.

the protocol: the fourteen-day audit.

this is what i would send you if you DMed me the screenshot of the flat line. it is not a new diet. it is fourteen days of finding out which of the four you have, because each one has a different fix and cutting calories is the right fix for exactly one of them, and even then only after the other three are ruled out. change nothing about your food targets for fourteen days. change how you measure.

so. is it me.

not in the way you think. it is not your metabolism, which in every trial of people like you was within a few percent of normal. it is not your hormones, which make the deficit harder to hold and do not stop fat from leaving. it is not that you ate too little, which costs muscle and not fat loss. it is, in order of how often the trials find it: the number in the app is lower than the number in you, the number on the watch is higher than the number in you, the number on the scale is mostly water this week, and the number monday to thursday was real and the number on the weekend was not. every one of those is fixable with a food scale, a weekly average, and two planned meals. none of them is fixable by eating less than you already are.

the version of me at 1,200 would have skipped this whole file and cut to 1,000. i know exactly what that costs now, and it was not fat. i am not going to tidy that up. i am going to tell you to buy a twelve-dollar food scale, weigh everything for fourteen days including the saturday, and stop eating back the watch. that is the whole intervention. the rest is reading a seven-day average and letting it be boring.

Quick answers

why am i not losing weight in a calorie deficit?

because in most measured cases the deficit is smaller than it looks, or the scale is hiding the loss. when researchers put people who reported eating under 1,200 calories a day and not losing into a lab for 14 days, their metabolism was within 5 percent of normal, they had under-counted food by 47 percent and over-counted exercise by 51 percent. add label error of about 18 percent on restaurant food, a wrist device that is off by 20 percent or more on calories burned, weekends that erase the weekday deficit, and three to four grams of water stored with every gram of glycogen, and a real deficit can read as a flat scale for weeks. audit those four before cutting calories again.

is starvation mode real? did eating too little break my metabolism?

the slowdown is real and small. after losing about 14 kilograms on 1,000 calories a day, resting metabolism ran 92 calories a day below what body size predicted, half that after four weeks at a stable weight, and 7 calories a day below prediction a year later. holding a 10 percent weight loss costs about 6 to 8 calories per kilogram of fat-free mass per day. that slows loss. it does not stop it or reverse it. the one famous case of a 500-calorie-a-day slowdown followed a 58-kilogram loss on a television show. what eating too little does reliably damage is muscle, not metabolism.

why does the scale go up on monday?

because weekends are measurably different days. in 80 adults weighed daily for up to 330 days, weight rose from saturday, peaked sunday and monday, and fell from tuesday. in a year-long trial, people who were in a real deficit monday to friday stopped losing on weekends, from higher intake on saturdays and lower activity on sundays. on top of the food, a carbohydrate-heavy weekend refills glycogen, which stores three to four grams of water per gram. monday's number is mostly water and a real weekend. the tuesday-to-friday trend is the number that means something.

how long should i wait before cutting calories again?

fourteen days of honest data, compared as weekly averages, not single mornings. the first week of a deficit drops water and glycogen fast, with a half-life under a week, and that phase runs four to six weeks before the loss becomes mostly fat, which is slower and easier to miss under a period, a hard leg day or a salty dinner. if the seven-day average is falling by 0.25 to 0.5 kilograms a week, nothing needs to change. if it is flat across two clean weeks with weighed food and weekends counted, cut by 150 to 200 calories or add a session, not both, and read another fourteen days.

keep reading

• why did i gain weight overnight. — the full water and glycogen arithmetic behind reason three
• why am i starving after the gym. — the compensation trials, and why half of people eat a session back
• why am i skinny and still soft. — recomposition, the thing the scale is worst at seeing
• how much protein do i actually need. — the 1.6 gram ceiling turned into meals
• why did i stop making progress. — the same four-week diagnosis logic, applied to the barbell

---

References

educational content, not medical advice. the numbers above are group results from small and mixed samples: the 1992 diet-resistance study is ten people, the sleep-and-fat trial is ten people over fourteen days, the protein-in-a-deficit trial is forty men over four weeks, the Minnesota-style restriction study is 32 men, and the weekend and metabolic-adaptation studies are 48 to 80 adults. the doubly-labelled-water datasets are large and observational. a flat scale alongside a missed period, hair loss, persistent cold or fatigue, or a pattern of restricting and rebounding is a conversation with a doctor or a registered dietitian, not a fourteen-day audit. everything in the tables is sourced.

- MoMo