RMR, TDEE, And Why Your Maintenance Calories Aren't What The Calculator Says

Almost every client who arrives at the practice has at some point used a TDEE calculator — Total Daily Energy Expenditure — to set their calorie target. Most of them are off by 200 to 500 calories per day. Some are off by more. The calculator isn’t being malicious. It’s using a formula that was never built for individual humans.

What TDEE is, and what it isn’t

Total Daily Energy Expenditure is the sum of four components: resting metabolic rate (the energy your body burns at rest, roughly 60–70% of the total), the thermic effect of food (energy used to digest meals, about 10%), non-exercise activity thermogenesis (NEAT — fidgeting, standing, walking around the day, 5–25%), and exercise activity (variable).

Online calculators estimate the first piece — RMR — using equations like Mifflin-St Jeor or Harris-Benedict, then multiply by an “activity factor” between 1.2 (sedentary) and 1.9 (very active). The equations are population averages built from cohorts of mostly young, mostly white, mostly Western subjects measured in the 1990s. They work surprisingly well on the median person and surprisingly poorly on most individuals.

What goes wrong

Müller et al. (2020) in the American Journal of Clinical Nutrition compared predicted RMR (via Mifflin-St Jeor) against measured RMR (via indirect calorimetry) in 1,150 adults. The equation was within 10% of the measured value for only about 60% of the cohort. For the other 40%, it under- or over-estimated by 200 to 500 calories per day. The errors clustered around two phenotypes: lean, muscular individuals (the equation under-estimates them) and adults over 50 who’d been on chronic caloric restriction (the equation over-estimates them, sometimes badly).

For a coaching protocol, a 300-calorie error is catastrophic. A client targeting a 300-calorie deficit, calculated from an estimate that’s already 300 calories too high, is actually eating maintenance. Twelve weeks of “dieting” produces no fat loss and a lot of frustration. The reverse error — under-eating by 300 calories without realizing — produces fatigue, irritability, and the metabolic slowdown the practice spends a lot of time un-doing.

What indirect calorimetry actually measures

Indirect calorimetry is the gold-standard RMR measurement. The client breathes into a mask for 10–15 minutes while resting; the device measures oxygen consumption and CO₂ production, and calculates energy expenditure from the gas exchange ratio. It gives an individual number, not a population estimate.

The Elite program at the practice uses a Breezing device for this, at baseline and follow-up. The numbers we see in practice diverge from calculator predictions about 35% of the time by more than 200 calories. Those are the clients whose protocols would have failed if we had trusted the calculator.

What this changes in the protocol

Three things, in coaching practice.

Targets are individual, not formulaic. Once we have a measured RMR, we add a measured NEAT estimate (we use a wearable for two weeks at baseline) and a known exercise expenditure to get a real TDEE. Caloric targets are set as percentages off that real number, not off a calculator.

Adaptation is monitored. RMR is not static. Sustained caloric deficits suppress it. Sustained over-feeding raises it. We re-measure at week 8 of any program to catch adaptive thermogenesis early and adjust the protocol before the client plateaus.

Macros are calibrated to the real number. Protein targets stay tied to lean mass (1.6–2.2 g/kg of lean body mass). Carbs and fats fill the remainder of a real caloric envelope, not a guessed one.

When the calculator is fine

For clients who can’t access indirect calorimetry, the practice uses a more honest workflow: start with the Mifflin-St Jeor estimate, then ADJUST based on 4 weeks of real weight and intake data. If weight tracking shows the estimate was wrong, we move the target by 100-calorie increments until the trend matches the goal.

This isn’t elegant. It is empirical. It treats the calculator as a starting hypothesis instead of a verdict.

The summary: calculators are useful for population averages and unreliable for individual people. Coaching that lives or dies on a number should measure that number, not estimate it.