In this guide
You've been walking after dinner for three months. You've logged every session, your watch says you've burned thousands of extra calories, and the scale hasn't budged. It's one of the most demoralising experiences in fitness, and it fuels a popular claim: that exercise barely burns any calories because the body just compensates. The truth is more interesting, and more useful, than either "exercise is useless for weight" or "every calorie burned counts".
A study published in Communications Medicine earlier this year, one of the most carefully measured of its kind, set out to see exactly where the energy goes when sedentary adults start exercising. It used doubly labelled water, the gold standard for measuring real-world energy expenditure, along with calorimetry and MRI. We read the full paper and translated it into what it means for your TDEE, your calorie targets and the temptation to "eat back" your workouts.
The short version: in 16 sedentary adults with overweight who did 12 weeks of supervised walking (about 200 minutes a week), total daily energy expenditure rose by about 194 kcal/day, close to the roughly 222 kcal/day the exercise itself cost. Yet weight barely changed (−0.2 kg), while fat mass fell by 1.2 kg and fat-free mass rose by 1.0 kg. Resting metabolic rate dipped by about 74 kcal/day, liver and kidney volumes shrank by about 5%, and compensation varied hugely: 11 of 16 people compensated, 5 didn't. Exercise isn't useless for fat loss. It's just less predictable than the calculators suggest.
The Claim That Exercise Doesn't Really Burn Calories
The idea comes from an influential line of research, notably by Herman Pontzer and colleagues, proposing that total daily energy expenditure is "constrained": when you add exercise, the body trims spending elsewhere, so the net increase is smaller than the exercise itself suggests. In its strongest form, the claim becomes "you can't out-exercise a bad diet" or even "exercise doesn't burn calories".
The opposing "additive" view says that exercise calories simply add on top of what you'd have burned anyway. Both views have evidence behind them, and the argument has been loud. What's been missing are tightly controlled human trials that measure each component of energy expenditure before and after training, which is what this paper tried to do.
The Study at a Glance
| Item | Details |
|---|---|
| Title | Multilevel metabolic adaptation to exercise training |
| Journal | Communications Medicine (Nature Portfolio), 2026; registered as NCT04460040 |
| Participants | 16 sedentary adults with overweight (half women; average age about 39, BMI about 28) |
| Intervention | 12 weeks of supervised aerobic walking, prescribed at about 20 kcal per kg of body weight per week (roughly 1,600 kcal/week) |
| Adherence | About 95%; around 3.9 sessions and roughly 200 minutes per week |
| Measurements | Doubly labelled water (total daily energy expenditure), metabolic chamber measurements, resting and sleeping metabolic rate, MRI of organ volumes, body composition, diet and physical activity monitoring |
| Extra arm | A mouse experiment examining liver tissue changes with exercise |
| Design note | Appears to be a single-group before-and-after study with no non-exercising control group |
What Participants Actually Did
The exercise dose was moderate and realistic: about four sessions a week, roughly 50 minutes each, mostly brisk walking, supervised so that the energy cost was known. The prescribed dose was set so that exercise would burn about 20 kcal per kg of body weight each week, which for these participants worked out at around 1,636 kcal per week, or roughly 222 kcal per day averaged across the whole week.
That's a sensible amount. It's the kind of exercise a health-conscious person might do, not elite training. It's also why the results speak to everyday readers.
The Scale Barely Moved. The Body Changed.
| Measure | Change over 12 weeks |
|---|---|
| Body weight | −0.2 ± 1.9 kg (not significant) |
| Fat mass | −1.2 kg (significant) |
| Fat-free mass | +1.0 kg (significant) |
| Visceral fat | Fell by roughly 10% |
| Fitness (VO₂max) | Rose by about 1.8 mL/kg/min |
This table is the single most important thing in the paper for anyone who weighs themselves daily. People lost fat, gained lean tissue, lost visceral fat and got fitter, and the scale told them almost nothing. If you judge exercise by weight alone, you can talk yourself out of a habit that's working. That's why we always suggest tracking waist, strength, photos and body fat, as in our Navy-method body-fat guide, alongside the scale.
The Energy Maths: Expected Versus Observed
| Quantity | Value |
|---|---|
| Baseline TDEE | About 2,508 ± 428 kcal/day |
| Cumulative energy cost of the exercise | About 18,465 kcal (roughly 222 kcal/day) |
| Weight loss predicted from that energy | About 2.4 kg (simple model) to 4.8 kg (dynamic model) |
| Weight loss observed | 0.2 kg, roughly 90–95% less than predicted |
| Change in TDEE after 12 weeks | +194 ± 304 kcal/day (significant) |
| Average "compensation" | About 97 ± 235 kcal/day, or roughly 39% of the exercise energy |
Here's the nuance. TDEE rose by about 194 kcal/day, which is most of the 222 kcal/day the exercise cost. So the "additive" view got most of the way there. But the body's energy stores fell by far less than a simple calculation predicts, and the authors call the difference "compensation".
What "Compensation" Actually Means
In this paper, compensation isn't a single biological mechanism. It's a bookkeeping term: the gap between the energy that exercise should have removed from the body's stores and the amount that actually went. That gap can come from changes in energy expenditure (the body spends less on other things) or changes in energy intake (you eat a bit more), and the study's method lumps them together.
That's why the "metabolism slowed down" explanation isn't the whole story. A real part of it is eating a little more without noticing, a little less incidental movement and a small dip in resting metabolism. Anyone who has felt unusually hungry after starting a new training programme will recognise it.
Where the "Missing" Calories Went
- Resting and sleeping metabolic rate fell. Resting metabolic rate dropped by about 74 ± 108 kcal/day, and sleeping metabolic rate fell too. The authors report that these decreases accounted for most of the measured compensation, at roughly 100 kcal/day.
- Organs got slightly smaller. MRI showed liver and kidney volumes down by about 5%, with brain volume unchanged. These organs burn a lot of energy for their size, and shrinking them may explain a fifth or so of the resting-metabolism dip.
- Walking became more efficient. Participants burned less energy for the same walking effort as they got fitter, so the planned calorie burn per session fell over time.
- Daily activity rose less than expected. Moderate-to-vigorous activity outside the sessions increased by smaller amounts than predicted, a sign that people unconsciously moved a little less elsewhere.
- Diet appeared roughly unchanged, as did the thermic effect of food. But self-reported intake is imprecise, so small increases wouldn't necessarily show up.
It's a "multilevel" story, as the title says. No single dial turned down. Many small ones did, a little each.
The Part That Matters Most: It Varied Enormously
Averages hide the most useful finding. Eleven of the 16 participants compensated for some of the exercise energy, and five didn't. Average compensation was 39% of the exercise energy, but the spread was huge (a standard deviation of 116 percentage points), meaning some people lost more than the exercise "should" burn while others lost almost nothing.
At the extreme, three participants showed a genuine fall in total daily energy expenditure of about 169 kcal/day, while the rest saw a rise of about 227 kcal/day. Compensation also tracked tightly with fat loss: the more someone compensated, the less fat they lost (a very strong correlation of 0.91). There was no clear link between the exercise "dose" and the amount of compensation.
For readers, that's the key message. Your body's response to exercise isn't fixed or predictable from a calculator. It's an individual pattern you can only learn by watching your own data.
Constrained Versus Additive: Where This Study Sits
| Model | Core idea | What this study found |
|---|---|---|
| Additive | Exercise calories add to TDEE one-for-one | TDEE rose by most of the exercise cost (194 of about 222 kcal/day) |
| Constrained | The body trims other spending to hold total expenditure steady | Resting and sleeping metabolism dipped by roughly 75–100 kcal/day, and some people compensated heavily |
| This study's picture | Neither extreme: a partial offset that varies by person | Total expenditure rose, but the body's energy stores fell far less than predicted |
That middle position is consistent with how many physiologists now think about it: exercise raises expenditure but not by as much as the calculators imply, and the offset differs from person to person. It also fits with how calorie targets are best treated: as starting estimates to refine.
What It Means for Your TDEE and "Eating Back" Calories
Three practical conclusions follow, and they're the opposite of "don't bother":
- Don't add exercise calories on top of your TDEE. The activity multiplier in a TDEE calculator already accounts for your typical week's movement. Using the TDEE calculator and then adding the calories from your watch double-counts. It also leans on wearable estimates that carry errors of 15–25% or more.
- Treat exercise calories as roughly half real, for planning. This is our rule of thumb, not a number from the paper. On average, about 61% of the exercise energy showed up in total expenditure here, and some people got less. If you must eat back anything, plan on half or less of what a tracker claims.
- Judge exercise by body composition and function. A scale that doesn't move can hide fat loss and lean-mass gain, as this study shows.
For a calculated estimate of what a particular session might cost, use the calories burned calculator, which uses MET values rather than heart-rate guesswork. Then remember that the number describes the session, not the lasting change in your daily budget.
A Three-Step Way to Calibrate Your Own Number
Because individuals differ so much, the most reliable way to know your TDEE is to measure your own response. It takes patience, but it works.
- Establish your baseline. For two weeks, eat a steady, tracked amount near your calculated maintenance (use the calorie calculator) and weigh yourself daily. Use the weekly average. If your weight is flat, that intake is your working maintenance.
- Change one thing. Add your new exercise habit without changing how you eat. Keep logging.
- Read the trend after 3–4 weeks. If your weekly average weight drifts down, you're in a deficit; if it holds, exercise is being offset. Adjust food by 100–200 kcal a day and repeat. Also check waist, strength and photos, because the scale may lag body composition.
The Limits You Should Know About
- Sixteen people. Small, mostly healthy sedentary adults with overweight; results may differ in leaner, older or more active groups.
- Twelve weeks of walking. Other exercise types, such as lifting or interval training, may behave differently.
- No separate control group. Without non-exercising comparators, some changes could reflect time, season or study participation.
- "Compensation" mixes intake and expenditure. Self-reported diet is imprecise, so the split between "eating more" and "spending less" isn't certain.
- Averages mask big individual differences. Most of the paper's own interest lies in the spread between people.
- Dose wasn't tested. The study didn't compare different exercise amounts, so it can't say whether more exercise leads to more or less compensation.
The Bottom Line
This careful trial doesn't say exercise is pointless for weight or fat loss. It says the body responds in layers: total expenditure rose by most of what the exercise cost, resting and sleeping metabolism dipped a little, organs got slightly smaller, daily activity rose less than expected, and the net effect on energy stores was smaller than a calculator predicts. It also showed that the scale can miss a favourable change in body composition entirely.
What it supports is a sensible, humble approach. Use a TDEE estimate as a starting point. Don't layer watch calories on top. Judge progress by waist, strength, fitness and weekly average weight, not a daily number. And expect your own response to differ from the textbook, which isn't failure. It's just physiology.
Frequently Asked Questions
Does exercise really burn the calories my watch says?
Not reliably. Wearable calorie estimates often have errors of 15–25% or more. And even true exercise calories don't translate one-for-one into stored energy loss. In a 2026 trial, total daily energy expenditure rose by about 194 kcal/day against about 222 kcal/day of exercise, yet body energy stores fell by far less than predicted.
Why didn't I lose weight when I started exercising?
Possibly because you lost fat and gained lean tissue, which can leave the scale unchanged, as in this trial (weight −0.2 kg, fat −1.2 kg, lean mass +1.0 kg). Another reason is partial compensation, from a slightly lower resting metabolism, less incidental movement or eating a bit more.
What is energy compensation?
The gap between the energy exercise should have removed from the body's stores and what actually went. It reflects both changes in energy expenditure and changes in intake. In the study, average compensation was about 97 kcal/day, or roughly 39% of the exercise energy, but it varied hugely between individuals.
Should I eat back my exercise calories?
Generally not on top of a TDEE-based target, since the activity level already accounts for your usual exercise. If you do add something, plan on about half or less of what a tracker claims, then check your weekly average weight against the plan.
Does exercise slow down my metabolism?
Slightly, in this trial. Resting metabolic rate fell by about 74 kcal/day after 12 weeks of walking, and liver and kidney volumes shrank by about 5%. Total daily expenditure still rose overall, so the net effect on calories burned was positive.
Why do some people compensate more than others?
The study found large individual differences: 11 of 16 compensated and 5 didn't. Greater compensation tracked with less fat loss, but the exercise dose didn't explain it. The reasons likely include appetite, fidgeting, organ and hormonal responses and genetics, though the study didn't pin them down.
How should I measure whether my exercise is working?
Track the weekly average of your weight, waist, strength and, if you like, body-fat estimates. Watch trends over three to four weeks rather than daily fluctuations. Fitness measures such as walking pace or heart rate recovery also show progress that weight can't.
Is this study proof that you can't out-exercise a bad diet?
It supports the idea that exercise alone often produces less weight loss than calculators predict. It isn't proof that exercise is pointless: participants still lost fat, gained lean mass, lost visceral fat and became fitter.
Does this apply to lifting weights or running?
Not directly. The trial used supervised walking only. Different exercise types may produce different compensation patterns, so these findings shouldn't be assumed to apply to everything.
References & Sources
- Multilevel metabolic adaptation to exercise training (2026). Communications Medicine. doi:10.1038/s43856-026-01502-z; open-access PMC version PMC13111656.
- ClinicalTrials.gov NCT04460040. Exercise-induced metabolic compensation: a physiological adaptive response to exercise training.
- Pontzer, H. et al. (2016). Constrained total energy expenditure and metabolic adaptation to physical activity in adult humans. Current Biology, 26(3), 410–417.
- Shcherbina, A. et al. (2017). Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort. Journal of Personalized Medicine, 7(2), 3.
- Ainsworth, B. E. et al. (2011). 2011 Compendium of Physical Activities. Medicine & Science in Sports & Exercise, 43(8), 1575–1581.
- MacrosMeasure (2026). Free TDEE Calculator.
- MacrosMeasure (2026). Free Calories Burned Calculator.
Calculate your own numbers — free.
18 science-backed calculators covering everything in this article and more.