Editorial note: The featured treadmill scene and inline energy graphic are AI editorial illustrations. They are not photographs, data figures, or measurements from the study discussed below.
Does the body compensate for exercise by spending less energy elsewhere? A 2025 study in Proceedings of the National Academy of Sciences found a positive, approximately linear relationship between habitual physical activity and total daily energy expenditure in its sample. The researchers did not find evidence that more-active participants had a fixed daily energy budget that simply shifted away from other functions. That result is interesting, but it is not proof that one additional walk causes a predictable number of extra calories to be burned in every person. The study was observational, and its authors explicitly state that causality cannot be established from these comparisons.
What the study actually tested
The paper’s correct title is “Physical activity is directly associated with total energy expenditure without evidence of constraint or compensation,” by Kristen R. Howard and colleagues. It appeared online in October 2025 in PNAS, with DOI 10.1073/pnas.2519626122. That matters because social captions sometimes substitute a punchier, nonexistent title. The published study asked whether people with higher habitual physical activity also had higher total energy expenditure, or whether expenditure flattened as activity increased. It also looked for possible behavioral or physiological signs of compensation.
Researchers recruited 75 adults from a region of Virginia. The group spanned sedentary people, recreationally active people, and endurance or ultra-endurance runners. Participants were 18 to 63 years old in the reported sample, and the investigators sought a broad range of running and walking volumes. The sample was not a random cross-section of all adults. People with several medical conditions, pregnancy, some medications, smoking, and other factors were excluded. Participants were selected to be weight-stable, a design choice that helps answer a narrower metabolic question but limits easy generalization to people actively losing weight.
Across roughly two weeks, the team measured total energy expenditure using doubly labeled water and assessed habitual movement using accelerometers. They also measured resting metabolic rate, body composition, and several blood-based markers. Doubly labeled water is a rigorous way to estimate energy used in daily life over days rather than just a laboratory workout, but it does not show a minute-by-minute movie of exactly where each calorie was spent. Accelerometers add movement information; neither method by itself converts the study into a randomized exercise intervention.
The two competing models
An additive model predicts that energy spent on movement generally raises the day’s total expenditure above what would have been spent at rest and on other activities. A constrained model predicts a point where additional activity is at least partly offset by reducing energy devoted to other processes, flattening the total. A compensation model can involve a partial offset even without a strict ceiling. These models have been debated because comparisons among populations, intervention studies, and extreme endurance settings do not always produce identical-looking results. The 2025 paper is one contribution to that debate, not the final word.
In this sample, the relationship between accelerometer-measured activity and total expenditure was best fit by a positive linear model, including after adjustment for fat-free mass. Resting metabolic rate did not show a significant association with activity in the reported analyses. Higher activity was associated with less sedentary time, rather than a simple pattern of active people moving less during the rest of the day. After correction for multiple testing, the investigators did not find the predicted associations between activity or energy expenditure and selected biomarkers of immune, reproductive, or thyroid function. Those findings collectively ran against the particular constraint and compensation patterns the team had specified.
There is an important statistical nuance: a relationship can be positive and still leave substantial individual variation unexplained. The paper reported an R-squared value around 0.35 for the unadjusted activity–total-expenditure model and around 0.37 after adjusting expenditure for fat-free mass. Those numbers do not mean that exercise only “works” one-third of the time. They mean the model explains only part of the variation among these people. Body size, diet, genetics, measurement noise, and other behaviors may matter. A group-level line is not a personalized calorie calculator.
What the result does not prove
The simplest overstatement is “exercise never causes compensation.” The researchers did not observe evidence of their tested forms of compensation in this weight-stable sample. They also said compensation could have occurred in components they did not measure directly. The thermic effect of food—the energy used to process food—was assumed rather than directly measured. Certain daily fluctuations in resting metabolism were not captured. Their high-activity participants were impressive, but the paper acknowledges that still-higher activity levels might behave differently. “No evidence here” is a careful statement; “impossible in humans” is not.
The second overstatement is that the experiment showed what happens when the same person starts exercising more. It did not. This was primarily a cross-sectional comparison of people who already had different habitual activity levels, observed during the measurement window. Such a design can identify associations and challenge a proposed model, but it cannot separate every characteristic that makes highly active people different from less-active people. An intervention following the same people through a sustained change in activity would answer a different causal question.
The third is that a higher energy expenditure guarantees weight loss. Body weight reflects energy intake as well as energy use, and appetite or eating behavior may change when someone becomes more active. The study’s participants were deliberately weight-stable; it was not a weight-loss trial. Even if increased movement adds to daily energy expenditure, it does not automatically prescribe how much a given person will lose or whether they should try to lose weight. Exercise is valuable for many outcomes beyond the scale, and turning this paper into a promised pounds-per-week formula would misstate its findings.
How this fits with earlier evidence
A frequently cited 2016 analysis by Pontzer and colleagues reported a constrained total-energy-expenditure pattern across activity levels. The newer PNAS study was designed partly to test that idea with detailed phenotyping and an activity range that included ultra-endurance runners. Different samples, methods, and conditions can yield different relationships. The strongest conclusion is not that one research group “destroyed” the other. It is that human energy allocation is complex and researchers are testing where different models apply.
The 2025 authors themselves discussed the possibility of energetic limits at levels beyond those observed in their sample. They also noted that energy balance matters: a person maintaining weight with adequate fueling may respond differently from someone in a sustained energy deficit. Thus, the paper gives evidence against a universal, readily reached fixed-energy ceiling under its study conditions. It does not establish that the body has unlimited capacity to raise expenditure or that all activity levels can be sustained indefinitely.
Useful advice without a calorie promise
For an individual reader, the practical message is to keep moving for health and capability, not to treat exercise as a precise transaction with a guaranteed burn. The U.S. Physical Activity Guidelines recommend that adults aim for 150 to 300 minutes of moderate-intensity aerobic activity weekly and muscle-strengthening work on at least two days. They also say that inactive adults should begin with small amounts and build gradually. These recommendations are based on a wider evidence base than this single study and do not depend on choosing one energy-budget model.
A brisk walk, a bike ride, or a manageable gym session can contribute to weekly activity. The right starting point depends on fitness, injury history, health conditions, and available time. Gradual increases help make movement sustainable, while strength work supports a broader range of physical tasks. If weight management is your aim, look at long-term patterns of activity, food intake, sleep, and health status rather than adding a watch’s estimated calorie figure to a diet plan as if it were exact. Wearables and formulas are approximations, not direct measurements of individual total expenditure.
The measured result is encouraging: in 75 weight-stable adults with varied habitual activity, more movement was associated with more total daily energy use, not with a detectable flattening within the observed range. The honest interpretation keeps the boundaries visible: a two-week observational measurement, a selected sample, and no causal guarantee for each new workout. It is a reason to take physical activity seriously, not a license to promise that every step permanently “boosts metabolism.”
One more useful distinction is between daily energy expenditure and resting metabolic rate. People often use “metabolism” to mean both, but the study measured total energy use across ordinary living as well as resting expenditure. Its headline association concerns the daily total. The absence of a significant relationship between activity and measured resting rate is part of why the authors did not see a simple resting-energy offset. It would be inaccurate to say that an active day permanently raises the body’s resting metabolic rate. The paper’s design was not built to establish that lasting adaptation.
Sources and study details
- Howard KR et al., PNAS, 2025, full text — study design, 75-person sample, measurements, model results, and limitations.
- PubMed record, PMID 41118225 — verified title, author list, date, DOI, and abstract.
- Pontzer et al., 2016, constrained expenditure analysis — earlier evidence discussed in the debate.
- U.S. Physical Activity Guidelines Q&A — adult activity recommendations and gradual-start advice.
