How this instrument works
Fat-Free Mass Index takes the weight that isn't fat — muscle, bone, organs, water, everything left once body fat percentage is subtracted out — and divides it by height squared, the same way body mass index divides total weight. VanItallie and colleagues proposed the measure in 1990 specifically because raw body weight conflates two very different things: how much of a person is fat, and how much is everything else. FFMI isolates the second figure and expresses it per unit of height, so a heavily muscled person and a person carrying excess fat at an identical body mass index no longer look alike on paper.
A further wrinkle showed up once researchers started using FFMI to study anabolic steroid use in strength athletes. Even after dividing by height squared, taller lifters still came out with slightly lower FFMI than shorter lifters carrying the same relative muscularity — an artifact of how lean mass scales with frame size. Kouri, Pope and colleagues published a correction in 1995: add 6.1 times the difference between 1.80 metres and actual height. Someone exactly 1.80 m tall receives no adjustment at all; shorter people gain a small bonus, taller people a small deduction, and the resulting normalized FFMI compares more fairly across lifters of very different builds.
None of this makes FFMI a medical instrument. It carries no diagnostic weight, no clinical reference range endorsed by a health authority, and no role in a doctor's assessment of anyone's health. Its real audience is bodybuilding and strength-training circles, where it works as a rough, back-of-envelope check on how much muscle a frame carries relative to its height — useful for tracking personal progress or weighing a claim, not for diagnosing anything.
- Enter Height in metres, centimetres, inches or feet — the equation converts internally to metres.
- Enter Weight in kilograms, pounds or stone.
- Enter Body fat (%) from calipers, a scan, or another estimate; FFMI is only as accurate as this number.
- Read Lean body mass, FFMI and Normalized FFMI; the working block shows each step of the arithmetic.
Worked example — 70 kg, 1.75 m, 20% body fat
Take a 70 kg man standing 1.75 m tall at 20% body fat. Lean mass first: 70 × (1 − 20 ⁄ 100) = 70 × 0.80 = 56 kg. Divide by height squared: 56 ⁄ 1.75² = 56 ⁄ 3.0625 = 18.29 — the raw FFMI. The height correction adds 6.1 × (1.80 − 1.75) = 6.1 × 0.05 = 0.305, giving 18.29 + 0.305 ≈ 18.59 for the normalized figure.
Push the same arithmetic further and the number gets interesting. A 100 kg, 1.70 m man at 25% body fat has 75 kg of lean mass, an FFMI of 75 ⁄ 1.70² = 25.95, and a normalized FFMI of about 26.56 once the +6.1 × (1.80 − 1.70) = 0.61 correction is added. That sits above the roughly-25 ceiling Kouri and colleagues reported for men who had never used anabolic steroids — not proof of anything about this particular person, but exactly the kind of figure the 1995 paper flagged as worth a second look.
Questions
What is a good FFMI score?
For adult men, the high teens to low twenties is typical with modest training experience, and the mid-twenties usually marks serious, sustained strength training. Kouri and colleagues found normalized FFMI in non-steroid-using men clustered under about 25, with very few natural lifters clearing that line — one reason a score near or above 25 draws extra scrutiny in sports-science writing, not because the figure alone proves anything about a given person.
Why does the formula add a height correction on top of dividing by height squared?
Dividing lean mass by height squared removes most, but not all, of the size effect — taller people still come out slightly underestimated relative to shorter people at the same true muscularity, an artifact of how skeletal frame and lean tissue scale together. Kouri's 1995 correction, 6.1 × (1.80 − height in metres), nudges the score back toward comparable footing, which is why FFMI and normalized FFMI are identical for anyone exactly 1.80 m tall.
Is FFMI a clinical measurement doctors use?
No. It has no standardized clinical cutoffs, no endorsement from a medical authority, and no role in diagnosing a health condition. FFMI's actual home is bodybuilding and strength-sport writing, where it works as an approximate, self-tracked gauge of muscularity relative to height rather than a figure a clinician would order.
Does a high FFMI mean someone uses steroids?
No — it is a statistical observation, not a test. Kouri and colleagues found that non-steroid-using men in their sample rarely posted a normalized FFMI above roughly 25, so a reading well past that line is unusual for natural muscle mass and invites questions. Genetics, limb proportions, hydration, and — most of all — how accurately body fat percentage was measured can all push an individual score around without any performance-enhancing drug involved.
How much does an inaccurate body fat reading affect FFMI?
A great deal, because body fat percentage sets lean mass directly: LBM = weight × (1 − body fat ⁄ 100). A caliper or scan reading that's off by five percentage points shifts lean mass, and therefore FFMI, by a proportional amount. Since body fat estimates themselves carry several points of error depending on method, treat any single FFMI figure as an estimate with a comparable margin, not an exact reading.
Does FFMI work the same way for women?
The arithmetic is identical, but the reference points built from Kouri's data were drawn mostly from men, so the commonly cited natural ceiling near 25 does not transfer directly to women, who carry a different typical ratio of lean to fat mass. Use FFMI to track a woman's own progress over time rather than to compare her figure against thresholds derived from a male sample.
References
- VanItallie et al. 1990, Am J Clin Nutr — introduced FFMI/FMI (PubMed)
- Kouri et al. 1995, Clin J Sport Med — height-normalization and steroid comparison (PubMed)
Read this first: This instrument computes a screening figure from population formulas — it is not a diagnosis, and it cannot see the whole picture a clinician can. Use it to inform a conversation, not to replace one.