Accuracy & limitations

These calculators are honest about being estimates. Here’s what that means in practice.

A consensus is agreement, not truth

The consensus is how closely the published formulas agree with each other — not how close they are to your real value. If four TDEE formulas cluster tightly, that’s reassuring, but they can all share the same bias. Treat the range (min–max) as the honest picture and the mean or median as a working estimate, not a measurement.

Some methods are rough by nature

• Body fat — the Navy (circumference) and Deurenberg (BMI-based) methods are population equations; expect a few points of error versus a DXA or hydrostatic measurement. Skinfold (Jackson-Pollock) needs careful caliper technique to be worth anything.
• RSMI wen-2011 — estimates appendicular muscle from height/weight/age/sex. It’s a screening estimate, materially less accurate than the direct DXA value. Prefer direct when you have a scan.
• Muscular potential — Casey Butt, the FFMI cap, and Berkhan are models of a natural ceiling, not guarantees; genetics vary widely around them.

Known scope limits

• The FFMI ≈ 25 natural ceiling comes from a male reference population (Kouri 1995). The practical limit for women is lower (~22); the above_natural_limit flag uses the single 25 threshold, so read it with that in mind.
• muscle-potential is men-only in v1 — it raises on sex: "female".
• Formulas are fit to populations. Individuals deviate. The further you are from a formula’s reference population (age, training status, ethnicity), the looser the estimate.

What the tools do guarantee

Determinism and validation. The same input always returns the same numbers, units are always explicit ({ value, unit }), and invalid input is rejected at the boundary — so the errors that remain are measurement and model error, not unit mix-ups or bad data. This is decision support, not medical advice.