Why BMI is wrong for athletes

The mechanics of the failure

BMI counts kilograms without asking what they're made of. Muscle is about 18% denser than fat, and trained athletes carry a lot of it: studies of professional and collegiate athletes routinely find half or more of a squad classified overweight by BMI while skinfold or DEXA testing shows body fat in the athletic range. The formula was validated on general populations whose extra kilograms were mostly fat. Feed it a population whose extra kilograms are mostly muscle and its core assumption collapses.

There is no correction factor that fixes this, because the information BMI would need (how the mass is distributed between muscle, fat, bone and fluid) is exactly the information it never collected. An index built from two numbers cannot recover a third. The only honest fix is to add a measurement that carries the missing signal, which in practice means a tape around the waist.

What the roster and squad studies actually found

The most-quoted demonstration is American football. Harp and Hecht applied BMI to all 2,168 players on National Football League rosters in the 2003–04 season and reported that 97% fell into the overweight or obese categories, with 56% meeting the BMI criterion for obesity. That is not a finding about the players; it is a finding about the instrument. A screen that flags virtually an entire occupational group is not discriminating between them.

The college data puts a number on the error rate. Ode and colleagues compared BMI against body fat measured by air displacement plethysmography in 226 athletes and 213 non-athletes. Sensitivity was high (BMI rarely missed someone who was genuinely overfat), but specificity in male athletes, female athletes and male non-athletes ran between 0.27 and 0.66. In plain terms, between about a third and three-quarters of the people BMI flagged as overfat were not. Female non-athletes showed the opposite pattern: specificity 0.90, sensitivity only 0.56, so BMI missed nearly half the women who genuinely carried high body fat.

Rugby is frequently cited in the same breath, and the pattern is plainly visible in published squad body-composition work, but we have not found a single roster-wide analysis for professional rugby equivalent to the NFL one, so we are not going to quote a percentage for it. The NFL and college figures above are what the literature actually supports, and they are more than enough to make the point.

Same BMI, different bodies

The table below makes the discordance visible. Every row is arithmetic, not a measurement of a real person: the heights, weights and waists are plausible for the group named, and the BMI, FFMI and waist-to-height figures are computed from them with a typical body fat percentage for that group applied. They are there to show the shape of the problem, not to be quoted as data.

Look at rows three and four in particular. Identical height, identical weight, identical BMI, and yet a body that differs by sixteen percentage points of fat, four and a half points of FFMI and a full ten centimetres of waist. BMI cannot tell them apart. Waist-to-height ratio separates them immediately, and FFMI shows you which one has been training.

Then look at the last two rows, which are the honest half of the story. Front-row forwards and offensive linemen carry genuinely high fat mass behind genuinely high muscle mass, and their waist-to-height ratios clear 0.5 comfortably. The discordance test does not excuse them, and it should not: that is the test working, not failing.

The discordance test

Here is the practical check the 2025 obesity framework formalised: a raised BMI only confirms excess body fat when a fat-distribution measure is raised with it. If your BMI is 28 but your waist-to-height ratio is under 0.5 and your waist sits below the WHO risk thresholds, the framework reads that as discordant: the pattern of muscle, not of visceral fat. Our BMI calculator runs this multi-measure check automatically when you add a waist measurement, and says so plainly when your numbers show the athlete pattern.

What makes this more than a convenient excuse is that the test is symmetrical, and the framework applies it in both directions. A normal BMI with a waist over half your height is also discordant, and that combination (sometimes called normal-weight central adiposity) carries real cardiometabolic risk that BMI alone waves straight through. The same logic that clears the lean 28 flags the sedentary 23. If you only accept the half of the result you like, you are not running the test.

What to track instead

Three measures serve a training person far better. Body fat percentage (even the tape-based Navy estimate, used consistently) tells you what the scale weight is made of. FFMI (fat-free mass index) tracks how much muscle you carry for your height, which is the number BMI garbles; it's the honest way to watch muscle gain over years. And waist-to-height ratio guards the one risk BMI was trying to flag in the first place, central fat, without being confused by your deadlift.

One caution keeps this honest: BMI's failure for athletes doesn't mean high BMI is harmless for everyone active. Carrying heavy mass (whatever it's made of) still loads joints, and some strength athletes at very high weights do carry substantial visceral fat behind the muscle. The discordance test cuts both ways: if the waist measures are elevated too, that's real, however strong you are.

• Waist-to-height ratio: one threshold at 0.5, no sex or age adjustment, and it moves when visceral fat moves.
• Body fat percentage from a tape or skinfold method: imprecise in absolute terms, reliable for tracking change if you measure the same way each time.
• FFMI: the muscle side of the ledger, and the number that shows years of training as a rising line.
• Performance and strength markers: the ones you actually train for, and the only measures on this list that are outcomes rather than proxies.
• Blood pressure, fasting glucose or HbA1c and a lipid panel: the things every anthropometric measure is ultimately a cheap stand-in for.

FFMI, and what a realistic ceiling looks like

Fat-free mass index is BMI with the fat taken out: work out your fat-free mass by subtracting your estimated fat mass from your body weight, then divide by height in metres squared. The recreational lifter in the table above (88 kg at 15% body fat and 1.78 m) has 74.8 kg of fat-free mass, and 74.8 ÷ 3.1684 gives an FFMI of 23.6. His sedentary counterpart at identical height and weight sits at 19.2. BMI called them the same.

The useful reference point comes from Kouri and colleagues, who measured fat-free mass index in drug-free and steroid-using athletes and found that drug-free lifters very rarely exceeded a normalised FFMI of about 25. That figure is a rough ceiling drawn from one study population, not a law, and it says nothing about any individual, but it does give scale. An FFMI in the low twenties represents years of consistent training; one in the high teens is untrained-to-average.

FFMI inherits the error of whatever body fat estimate you feed it, so treat the absolute figure with the same suspicion you would give that estimate, a couple of points either way. What it does well is track: measured the same way every few months, a rising FFMI at stable body fat is unambiguous evidence that the extra kilograms on the scale are muscle, which is exactly the claim BMI cannot evaluate.

Where the athlete exemption stops

The discordance argument is frequently stretched past where the evidence supports it. It applies to people whose waist-based measures are genuinely below their thresholds. It does not apply simply because someone lifts, plays a contact sport, or has a physical job. If your waist-to-height ratio is 0.55, you do not have the athlete pattern: you have a raised BMI and raised central adiposity, and the fact that there is a lot of muscle underneath does not subtract the fat on top.

Certain positions and weight classes live in that territory by design. Front-row forwards, offensive and defensive linemen, heavyweight throwers and super-heavyweight strength athletes carry mass deliberately, and a substantial part of it is fat. The Harp and Hecht data bears that out: 56% of NFL players over BMI 30 is not an artefact spread evenly across the roster. For those athletes the relevant questions are about blood pressure, glucose and lipids during the career and about weight trajectory after it, not about whether a category label is unfair.

There is also a purely mechanical point that no body composition measure addresses. Joint loading scales with total mass, not with the fat fraction of it. A 120 kg athlete loads a knee like 120 kg regardless of how lean those kilograms are, which is why heavy athletes carry elevated osteoarthritis risk whatever their body fat percentage reads. That is a training and career-management issue rather than a metabolic one, but it is a real cost of carrying the mass.

What to say at a medical check

Arriving with numbers changes the conversation. Bring your waist measurement and your waist-to-height ratio, a body fat estimate and how you arrived at it, your training history in plain terms (what you lift or play, how many sessions a week, for how many years) and your weight trend over the last year rather than a single reading. All of that is information a clinician can use; a complaint that BMI is unfair is not.

Ask specifically that the waist measurement is recorded alongside the BMI. Since January 2025 that is what the criteria call for, so it is a reasonable request rather than a special favour, and it puts the discordance on the record where it will still be visible at the next appointment. If a BMI-based threshold is being used to gate something (a procedure, an insurance rating, an occupational assessment), ask which classification system and which criteria are being applied, because those decisions often still run on BMI alone.

Then let the blood work settle it. Blood pressure, fasting glucose or HbA1c, and a lipid panel measure the outcomes that every anthropometric index is a cheap proxy for. If those are clean and your waist measures sit below their thresholds, the case that a raised BMI is an artefact of muscle is about as strong as it can be made without a DEXA scan. If they are not clean, that is worth knowing far more than the label was.

Frequently asked questions

Put it into practice

Run your own numbers through the FFMI calculator, the body fat calculator, the waist-to-height calculator and the BMI calculator. Related reading: BMI explained, BMI vs body fat vs waist and Navy formula accuracy.

Sources

1. Harp JB, Hecht L. Obesity in the National Football League. JAMA 2005;293:1061–2. doi.org/10.1001/jama.293.9.1061-b
2. Ode JJ, Pivarnik JM, Reeves MJ, Knous JL. Body mass index as a predictor of percent fat in college athletes and nonathletes. Med Sci Sports Exerc 2007;39:403–9. doi.org/10.1249/01.mss.0000247008.19127.3e
3. Kouri EM, et al. Fat-free mass index in users and nonusers of anabolic-androgenic steroids. Clin J Sport Med 1995;5:223–8. doi.org/10.1097/00042752-199510000-00003
4. Rubino F, et al. Definition and diagnostic criteria of clinical obesity. Lancet Diabetes Endocrinol 2025;13:221–262. doi.org/10.1016/S2213-8587(24)00316-4
5. Ashwell M, Gibson S. Waist-to-height ratio as an indicator of early health risk. BMJ Open 2016;6:e010159. doi.org/10.1136/bmjopen-2015-010159
6. WHO. Waist circumference and waist–hip ratio: report of a WHO expert consultation. Geneva, 2008. www.who.int/publications/i/item/9789241501491