Energy compensation and why exercise calories may not add up

Studies without constraint deserve the same attention

A 2025 observational study found that greater activity was associated with greater total expenditure, without evidence of constraint or compensation. The positive relationship was linear both before and after adjustment for fat-free mass, meaning the mass of the body excluding fat. People with higher activity also spent less time sedentary. The findings did not support a plateau in total expenditure as activity increased.[3]

The study also examined indicators of immune, reproductive and thyroid function. After accounting for the possibility of chance associations across multiple comparisons, it reported no association of the adjusted markers with activity, total expenditure or resting metabolic rate. The physiological changes one might expect from a constrained model were not confirmed by those measurements.[3]

These are findings against constraint under the conditions examined, rather than an experimental rejection of every possible compensatory response. The design was observational. Just as an observational plateau cannot prove what happens when a particular person starts exercising, an observational linear relationship cannot describe every individual response to an assigned exercise program. Applying the same standard to both sides keeps the comparison fair.[3][9]

A short intervention provides another challenge to the constrained model. In a repeated-measures experiment, 12 adults completed a high-activity period and a low-activity period, each lasting 10 days. Activity expenditure increased by about 28%, or approximately 250 kcal per day. Total expenditure rose by 10%, or 272 kcal per day. Resting expenditure adjusted for body size did not decline.[7]

Within that period and those conditions, the findings did not match the prediction that a saving at rest would offset the extra activity. The increase in the whole-day total was close to the increase in activity expenditure. The study therefore supplied a different result from the interventions in which total expenditure rose by less than expected.[7]

The same paper analyzed cross-sectional data from 268 adults. A cross-sectional analysis compares different people at a given point, rather than following an assigned change over time. In that dataset, size-adjusted resting expenditure and activity expenditure were statistically independent. Combining that result with the short repeated-measures experiment supported the additive model.[7]

The time limit still matters. A result from short activity periods does not establish that compensation is absent throughout a long exercise habit. Conversely, findings from longer interventions cannot be inserted into this short experiment as though the participants had shown them. Each study answers a question within its own population, duration and measurements.[2][4][7]

Reading only the studies supporting constraint can make a plateau seem inevitable. Reading only the studies supporting addition can make every extra activity calorie seem fully available in the daily budget. Neither approach resolves the conditions under which the different patterns appear. The evidence leaves a task of distinguishing when compensation occurs and when changes look more nearly additive.[1][7][8]

For someone whose weight has stayed steady despite exercise, this debate should not become a judgment of effort. It asks which explanation fits measured expenditure, not whether the person deserves credit for exercising. A scale reading alone cannot place you inside either model. It does not separate total expenditure from resting expenditure, and it does not establish the size of any shortfall from the exercise prescription.[2][7]

Match the duration and calculation before comparing rates

A systematic review of predictors of compensation included 61 studies and 928 participants. Average compensation was 18% ± 93%. The spread was larger than the average, which makes a uniform rate difficult to apply. The review defined compensation through changes in body energy, including fat mass and fat-free mass, relative to the total energy expended through exercise.[19]

This definition is essential. The review was asking how changes in energy stored in the body compared with the exercise expenditure. Other papers ask how much total daily expenditure rose compared with the increase predicted from activity. Both questions relate to disappointing weight-loss calculations, but their percentages are not interchangeable. A shared label does not make the underlying measurements identical.[1][2][19]

The interaction between initial fat mass, age and intervention duration explained 48% of the variation in compensation in the review. Sex, exercise frequency, intensity and exercise energy expenditure were not significant predictors. The result therefore cannot be reduced to “more exercise always produces more compensation.” It was an analysis involving combinations of characteristics and the length of the intervention.[19]

The fitted model described different patterns by duration. In shorter studies, younger people with higher initial fat mass were estimated to have lower compensation, while younger people with lower initial fat mass had higher compensation. From 25 weeks onward, compensation no longer differed by those predictors. In studies lasting about 80 weeks, compensation approached 84%.[19]

That 84% is not an instruction to remove 84% of the calories on today’s exercise display. It came from a model using body energy changes and incorporating duration and predictors. The shorter-term differences involving age and initial fat mass did not remain in the same way over the longer term. Keeping the definition and duration attached to the number is part of understanding it.[19]

Nor does greater compensation in longer interventions become a reason to stop exercising. The review compared anticipated energy deficits with changes in body energy stores. It did not measure every reason someone might value an exercise habit. The walking intervention, for example, reported improved body composition despite minimal weight loss. A decision about continuing a routine should keep its actual content separate from a prediction of weight change.[4][19]

When comparing papers, first identify the denominator: is it prescribed exercise expenditure or the measured increase in activity expenditure? Then identify the numerator: is it the rise in total expenditure or the change in stored body energy? A study showing a smaller rise in expenditure and a study showing a smaller loss of stored energy can illuminate related concerns without measuring the same outcome.[1][2][19]

Adjustment for body size and composition is another part of the comparison. A review of models, measurements, experimental designs and statistical methods found that, after accounting for spurious correlations, the available longitudinal and cross-sectional evidence most supported the additive model. It also called for more and better data. An apparent correlation should not automatically become a claim that activity caused expenditure elsewhere to fall.[8]

For a practical reading note, keep the population, measurement and duration together. An average across adults living ordinary lives, a small supervised walking intervention and a short experiment changing activity each cover a different range of questions. Keeping those details together makes it clearer that no research percentage can serve as a personal extra food allowance.[4][5][7]

This approach also helps when advice seems contradictory. Before choosing the more reassuring number, ask whether the advice has kept the study’s actual outcome attached. A percentage of a predicted rise, a percentage compensated, and a percentage based on body energy changes can sound like rival estimates of the same thing. Often the first problem to resolve is that they describe different things.[1][5][19]

Common misconceptions

A fixed correction to exercise calories is not established as a way to make personal calculations accurate. The analysis reporting average compensation of 28% also described substantial differences associated with body composition. The systematic review found a very wide spread of responses. A population estimate cannot simply become your daily correction factor.[5][19]

The assumption that compensation means exercise never increases expenditure is also too strong. Total expenditure increased in the small walking intervention supporting constraint. In the short activity experiment, its increase was close to the increase in activity expenditure. The disagreement concerns the size and conditions of the response, and it includes results between the extremes of complete addition and no increase.[4][7]

A further misconception is that limited weight loss proves metabolic adaptation. The 24-week trial identified participants with compensation but did not find significant group differences in adaptation during rest or sleep. The walking intervention did find reductions in those metabolic rates. A change in your body weight cannot diagnose which of these pathways occurred.[2][4]

Feeling tired after a session does not resolve the question either. “I feel exhausted” and “my expenditure elsewhere fell” are different observations. The studies assessed activity and metabolism rather than calculating compensation from fatigue alone. Keep how you felt separate from what you measured; that leaves room to describe the experience without assigning it a mechanism the record cannot establish.[2][4][7]

What you can do today

These are practical planning suggestions, not tested prescriptions for commuting, combining exercise types or delaying food cuts.

• Choose a walkable part of your usual travel and walk it. When planning how to follow the habit, use the study’s 12-week supervised walking program as context; its target of 20 kcal/kg/week is a research dose, not a personal prescription.[4]
• Include manageable resistance training as an option alongside aerobic activity in your next exercise plan. The comparison suggested that compensation may be smaller with resistance training, but established no frequency for this combination.[1]
• Eat your planned meal after exercise without automatically adding a plate or snack to match the workout display. This suggestion does not call for reducing your intake.
• If you are restricting food and adding exercise, pause before cutting food further because weight loss missed a prediction. The comparison suggested greater compensation when aerobic exercise accompanied dietary restriction.[1]
• Record the exercise completed and the food actually eaten. Review weight change separately from whether you carried out the activity, without guessing a daily compensation rate.

Plan movement beyond calorie arithmetic at On the Shore Tachikawa

When exercise-calorie calculations leave you unsure what to do next, choose an activity you want to attend at On the Shore Tachikawa (オンザショア立川店) and compare its opening hours with your schedule.

Many of our yoga instructors speak English, and the studio’s owner often helps guests from the US bases in English herself. Pregnant guests cannot join lava-stone hot yoga; room-temperature maternity yoga is available. Anyone whose doctor has advised against exercise cannot participate.

The studio is on 3F of the Etoile Bldg, 2-14-10 Akebonocho, Tachikawa, Tokyo, a 1-minute walk from JR Tachikawa Station’s North Exit. Opening hours are 8:00–23:30 every day.

Its lava-stone hot yoga uses a studio floor laid with lava-stone plates. Other options include room-temperature yoga, Pilates, women-only kickboxercise, boxercise, HIIT and personal training. Yoga alone includes more than 25 kinds of lessons.

The yoga trial is a 60-minute lesson costing ¥1,980 including tax, with 2 bath towels, 1 face towel and yoga mat rental included. The studio website describes the studio; prices and trial booking are available through these links.

References

1. The evidence for constrained total energy expenditure in humans and other animals. — H. Pontzer et al., 2026, Current biology : CB. DOI: 10.1016/j.cub.2026.01.025
2. No evidence for metabolic adaptation during exercise-related energy compensation — E. W. Flanagan et al., 2024, iScience. DOI: 10.1016/j.isci.2024.109842
3. Physical activity is directly associated with total energy expenditure without evidence of constraint or compensation — Kristen R. Howard et al., 2025, Proceedings of the National Academy of Sciences of the United States of America. DOI: 10.1073/pnas.2519626122
4. Multilevel metabolic adaptation to exercise training — T. Knaan et al., 2026, Communications Medicine. DOI: 10.1038/s43856-026-01502-z
5. Energy compensation and adiposity in humans — V. Careau et al., 2021, Current biology : CB. DOI: 10.1016/j.cub.2021.08.016
6. Longitudinal and cross‐sectional evidence that daily resting and activity energy expenditures are independent in humans — Andrew K. Yegian et al., 2026, The Journal of Physiology. DOI: 10.1113/jp291108
7. Physical activity and metabolic rates in humans. — Daniel E. Lieberman et al., 2026, The Journal of experimental biology. DOI: 10.1242/jeb.251083
8. Constrained Total Energy Expenditure and Metabolic Adaptation to Physical Activity in Adult Humans — H. Pontzer et al., 2016, Current biology : CB. DOI: 10.1016/j.cub.2015.12.046
9. Predictors of Energy Compensation during Exercise Interventions: A Systematic Review — Marie-Ève Riou et al., 2015, Nutrients. DOI: 10.3390/nu7053677

Cover photo: A woman rests on a yoga mat, eyes closed, with one arm across her forehead. (Photo: Shixart1985 / CC BY 2.0 / Wikimedia Commons)

Information as of October 2026. For your own health, consult a doctor.