A BMI of 27 sounds like a verdict. The label “overweight” gets attached, the patient gets a brochure, and the appointment moves on. But that single number — body mass divided by height squared — cannot tell you whether the person sitting across from you is a powerlifter, a sedentary office worker, a postmenopausal woman with bone loss, or a 70-year-old of South Asian ancestry whose actual cardiometabolic risk is being seriously underestimated. Body mass index is a useful tool. It is not a diagnosis. This article explains what BMI was designed to do, what the standard categories mean, and the places where treating the number too literally goes wrong.
What BMI is and where it came from
Body mass index is defined as mass in kilograms divided by height in meters squared:
- BMI = mass (kg) / height (m)2
- Or equivalently in U.S. customary units: BMI = 703 × mass (lb) / height (in)2
The formula is the work of Adolphe Quetelet, a Belgian astronomer and statistician, who introduced it in the 1830s as part of his investigation into the statistical properties of the “average man” (l'homme moyen). He was studying the distribution of body weight across populations and looking for a way to normalize weight by stature so that adults of different heights could be compared. He never proposed it as a clinical measure of individual health. The term “body mass index” itself was not coined until 1972, when the physiologist Ancel Keys argued that Quetelet's ratio correlated reasonably well with body fat estimated from skinfold thickness in large samples — and was simpler and cheaper than any alternative.
That history matters. BMI is fundamentally a population-level statistic that has been borrowed for individual screening. It survives in clinical practice not because it is the best single measure of health risk, but because it costs nothing to compute and travels easily across charts, registries, and epidemiological studies.
The standard WHO and CDC categories
The cutoffs most clinicians and consumer apps use are the WHO adult classifications, adopted by the CDC and the NIH for U.S. clinical guidelines.
| Category | BMI range (kg/m2) | Notes |
|---|---|---|
| Underweight | < 18.5 | WHO sub-classifies as mild (17.0–18.49), moderate (16.0–16.99), severe (< 16.0). |
| Normal weight | 18.5 – 24.9 | Sometimes called “healthy weight range.” |
| Overweight (pre-obese) | 25.0 – 29.9 | WHO terminology emphasizes this is a screening threshold, not a disease. |
| Obesity class I | 30.0 – 34.9 | Increased risk of comorbidities. |
| Obesity class II | 35.0 – 39.9 | Severely increased risk. |
| Obesity class III | ≥ 40.0 | Sometimes called “severe” or “morbid” obesity in older literature. |
These categories are clinically useful in aggregate. The relative risk of type 2 diabetes, hypertension, certain cancers, and all-cause mortality really does climb with BMI in large populations. The problem is what those bands obscure when applied to a single individual — and it is a substantial problem.
Mistake 1: treating BMI as a measure of body fat
BMI does not distinguish lean mass from fat mass. The numerator is total body weight, which includes muscle, bone, organs, water, and adipose tissue. Two people with identical BMIs can have vastly different body compositions, and therefore vastly different metabolic and cardiovascular risk profiles.
The classical example is a muscular athlete. A 1.83 m (6 ft) rugby player weighing 100 kg has a BMI of about 29.9 — squarely in the “overweight” category, just below the obesity threshold. His actual body-fat percentage might be 12%. Conversely, an older adult who has lost lean mass (a phenomenon called sarcopenia) can sit at a “normal” BMI of 22 while carrying a body-fat percentage above 35% — sometimes called “normal-weight obesity,” or sarcopenic obesity when muscle loss is the dominant mechanism. This pattern is associated with elevated metabolic risk despite a reassuring number on the chart.
The CDC's own guidance is explicit: BMI is a screening tool, not a diagnostic of body fatness or health. A clinician should follow up an out-of-range BMI with additional assessment, not act on the number alone.
Mistake 2: ignoring age, sex, and ethnicity
BMI uses one set of cutoffs for all adults, but the relationship between BMI and body fat varies systematically by group.
- Sex. At any given BMI, women on average carry more body fat than men, because men have a higher proportion of muscle mass. The standard cutoffs partially smooth over this difference but do not fully account for it.
- Age. Older adults tend to lose lean mass and gain adipose mass even when total body weight stays constant. A BMI of 24 at age 30 and a BMI of 24 at age 75 do not represent the same body composition or the same risk profile.
- Ethnicity. The 2004 WHO Expert Consultation, published in The Lancet, reviewed evidence that adults of South Asian, East Asian, and Southeast Asian ancestry develop adverse cardiometabolic risk (type 2 diabetes, dyslipidemia, hypertension) at lower BMIs than European-ancestry populations. They proposed lower public-health action thresholds: roughly BMI ≥ 23 as overweight and BMI ≥ 27.5 as high-risk for many Asian populations. These cutoffs are now used in clinical practice in India, Singapore, China, Japan, and elsewhere. Conversely, people of Black African and Afro-Caribbean ancestry tend to have higher lean mass and lower visceral adiposity at a given BMI than European-ancestry adults, so the standard cutoffs may overestimate fat-related risk somewhat in those groups.
- Pregnancy. BMI is not meaningful during pregnancy. Pre-pregnancy BMI is what is used to guide gestational weight-gain recommendations.
- Children and adolescents. Adult cutoffs do not apply. Pediatric BMI uses age- and sex-specific percentile charts (CDC and WHO each publish growth standards), not the 18.5 / 25 / 30 thresholds.
Mistake 3: ignoring fat distribution
Where adipose tissue is stored matters more for cardiovascular and metabolic risk than how much of it there is. Visceral adipose tissue — fat around the abdominal organs — is metabolically active and independently associated with insulin resistance, dyslipidemia, and cardiovascular events. Subcutaneous fat on the hips and thighs is much less so. BMI cannot see the difference.
Two simple, low-cost measures do better:
- Waist circumference.WHO action thresholds (substantially increased risk) are typically > 102 cm (40 in) for men and > 88 cm (35 in) for women, with lower thresholds recommended for some Asian populations. Waist circumference adds predictive information beyond BMI for cardiovascular risk.
- Waist-to-hip ratio.WHO cutoffs of > 0.90 for men and > 0.85 for women are associated with elevated risk. Several large prospective studies have found waist-to-hip ratio to be a stronger single predictor of myocardial infarction risk than BMI.
The 2008 WHO expert consultation on waist circumference and waist-to-hip ratio recommended that these be used alongside BMI rather than instead of it, because they capture complementary information. A patient with a normal BMI but a high waist circumference (sometimes called “TOFI” — thin outside, fat inside) is exactly the kind of person who would be missed by BMI screening alone.
What BMI is genuinely useful for
After all those caveats, BMI still earns its place — for specific purposes.
- Population-level epidemiology. BMI is cheap, reproducible, and well-validated for comparing prevalence of underweight and obesity across regions and over time. The WHO uses it for global surveillance for exactly these reasons.
- Initial screening at the population or clinical level. A BMI well outside the normal range is a reasonable trigger for further assessment — a conversation, a waist measurement, a lipid panel, an A1c — rather than a starting point for treatment.
- Eligibility and dosing in some clinical contexts. Bariatric surgery guidelines, anesthesia planning, and certain drug-dosing protocols use BMI thresholds as practical entry criteria.
- Tracking change over time within an individual. Even though absolute BMI is a noisy measure of fatness, large intentional changes in BMI within the same person usually reflect real changes in adiposity and are easy to monitor.
Worked example: same BMI, very different bodies
Consider three adults, each with a calculated BMI of exactly 27.0 — the middle of the “overweight” range.
- Patient A: 1.80 m, 87.5 kg, age 28, male, college rower. Body-fat percentage by DEXA: 14%. Waist 84 cm. Resting blood pressure 118/72. A1c 5.1%. The BMI category is misleading; his cardiovascular risk profile is excellent and the elevated weight is overwhelmingly lean mass.
- Patient B: 1.65 m, 73.5 kg, age 56, female, sedentary, post-menopausal. Body-fat percentage 38%. Waist 96 cm (above the WHO threshold of 88 cm). Fasting glucose 108 mg/dL, A1c 5.9%. Same BMI, but the waist circumference and labs make the cardiometabolic risk much more concrete than the BMI category suggests.
- Patient C:1.70 m, 78 kg, age 45, male, of South Asian ancestry. Under WHO standard cutoffs, “overweight.” Under the lower Asian cutoffs (overweight ≥ 23, high-risk ≥ 27.5), he sits essentially at the high-risk action threshold. Combined with a waist circumference of 94 cm — already above the lower ethnic-specific cutoff often used in clinical practice (> 90 cm for Asian men) — the picture is meaningfully different than the generic chart suggests.
Three identical BMIs, three very different conversations. None of these conclusions could be reached from the BMI alone.
Common pitfalls in interpretation
- Treating category boundaries as cliffs.The difference between BMI 24.9 (“normal”) and BMI 25.0 (“overweight”) is one tenth of a unit, well within measurement noise from a single weighing. The categories are convenient labels, not biological transitions.
- Using a single measurement. Body weight fluctuates 1–2 kg day to day from hydration, glycogen, and gut contents. Averaged measurements over a week or comparison of trends matter more than any single value.
- Applying adult cutoffs to children, athletes, the frail elderly, or pregnant patients. All four groups require different reference standards or different tools entirely.
- Confusing BMI with body-fat percentage. They correlate, but the correlation is loose and group-dependent. Equipment that estimates body-fat percentage (DEXA, hydrostatic weighing, multi-frequency bioelectrical impedance) gives different information than BMI.
- Assuming “normal” BMI means no cardiometabolic risk. Normal-weight individuals with elevated waist circumference, sedentary lifestyles, or family history can carry substantial risk. BMI is not a replacement for standard screening labs and history.
- Reporting BMI without units or self-reported height adjustments. Self-reported heights tend to be inflated and self-reported weights deflated, which biases self-calculated BMI downward. Measured values are preferred where decisions hinge on them.
The takeaway
BMI is a 19th-century population statistic that has had a successful, if imperfect, second career as a clinical screening tool. It is fast, free, reproducible, and roughly correlated with risk in large groups. But it cannot tell muscle from fat, it ignores where fat is stored, and the standard cutoffs were derived in populations that do not represent everyone they are now applied to. For an individual patient, BMI should prompt a conversation, not end one — and that conversation is much better informed by waist circumference, family and personal history, activity level, and labs than by the BMI band alone. Use the number for what it is: a starting point.
This article is for educational purposes and does not constitute medical advice. Decisions about weight, body composition, and cardiometabolic health should be made in consultation with a qualified clinician.