Explainer - Published August 20, 2026
Most disagreements come from different questions hidden behind the same “calorie calculator” label: resting energy or total energy, one formula or another, a selected activity band, a body-fat estimate, and a static or dynamic forecast.
The practical answer
BMR estimates energy used at rest. TDEE multiplies that value to represent the rest of the day. A BMR result should be lower than a TDEE result for the same profile.
Mifflin-St Jeor and revised Harris-Benedict use different coefficients. In our benchmark they differed by 36 to 246 kcal/day after the same activity factor.
For a 1,600 kcal BMR, moving from factor 1.375 to 1.55 adds 280 kcal/day. Two sites can use the same equation and disagree because their labels lead users to different bands.
Katch-McArdle uses estimated lean mass. That can add useful information when body fat is measured well, but measurement error moves directly into the result.
Rounding, minimum intakes, deficit caps, macro assumptions, and static versus dynamic timelines all change the final recommendation even when maintenance starts at the same number.
Compare books for the Mifflin-St Jeor, Harris-Benedict, and Katch-McArdle methods
Deficit Compass tested 12 synthetic adult profiles with Mifflin-St Jeor TDEE, revised Harris-Benedict TDEE, weight x 10, and weight x 15. The two equations differed by 115 kcal/day on average. The full four-method range averaged 989 kcal/day.
The weight-only shortcuts created most of the spread because they ignore age, height, sex, and selected activity. That finding changed the product: Deficit Compass no longer gives those shortcuts an equal vote in the recommended maintenance value.
| Situation | Give more weight to | Keep visible | Next action |
|---|---|---|---|
| No body-fat measurement | Mifflin-St Jeor TDEE | Adjacent activity band and Harris comparison | Start, then calibrate. |
| Reliable recent body-fat measurement | Compare Katch-McArdle with Mifflin | Measurement uncertainty | Do not assume Katch is automatically correct. |
| Between activity bands | Lower band as a conservative start | Upper band as range boundary | Use two comparable weeks to test. |
| Consistent 2-4 week record | Trend-implied maintenance | Formula as reasonableness check | Adjust one variable at a time. |
| Large early weight change | More observation | Water, glycogen, sodium, and logging changes | Do not force a maintenance estimate yet. |
An arithmetic mean reduces the influence of any one estimate only when the estimates are credible measures of the same quantity. A detailed equation and a weight-only shortcut do not use equivalent information. Averaging them can pull the result toward a shortcut that is especially unsuitable for a particular age, height, sex, or activity pattern.
In benchmark profile P09, a sedentary 45-year-old man at 250 lb, Mifflin TDEE was 2,411 kcal and revised Harris-Benedict TDEE was 2,632. The weight x 15 shortcut was 3,750. Their four-model average was 2,823, more than 400 kcal above Mifflin.
That average is calculated correctly. The problem is interpretation: four numbers do not become four independent measurements merely because they are displayed together.
Predictive equations were built from groups, not from a metabolic measurement of the individual using the page. Activity multipliers compress a varied week into one category. Food logs and wearable estimates also contain error.
The right expectation is a useful starting range. If a consistent record shows that body weight is stable near 2,300 kcal/day, that evidence can be more useful for the next planning decision than debating whether a formula says 2,250 or 2,420.
None of these equations can account fully for pregnancy, breastfeeding, eating-disorder history, medication effects, illness, or individual clinical needs. If the estimated target is unusually low, symptoms are concerning, or weight management intersects with medical care, use qualified guidance rather than searching for the calculator that returns the preferred answer.
Written by Builderfox LLC from the published equations and the Deficit Compass benchmark dataset. This page has not yet been medically reviewed. Report an error through Contact.
Creation note: AI-assisted research and drafting were manually checked against the linked sources and data. See the editorial policy.
Use the rubric