BMI

BMI(Body Mass Index) is a single number summarising the relationship between someone’s weight and height. It’s computed as weight ÷ height²in kilograms and meters. The imperial form multiplies by 703: BMI = 703 × weight (lb) ÷ height (in)².

Devised by Belgian statistician Adolphe Quetelet in the 1830s and renamed BMI by physiologist Ancel Keys in 1972. The World Health Organization uses BMI as the primary population-screening tool for under- and over-weight, though it’s well known to misclassify individuals with atypical body composition (athletes, the elderly, children).

Standard WHO category bands: < 18.5 underweight, 18.5–24.9 healthy, 25.0–29.9 overweight, ≥ 30.0 obese (subdivided into classes I-III).

See our BMI calculator for the live tool, or our guide on BMI accuracy for when it misleads.

Worked example

A person 178 cm tall (1.78 m) weighing 82 kg has BMI = 82 ÷ (1.78)² = 82 ÷ 3.1684 = 25.9, landing in the “overweight” band. In imperial: 5′10″ (70 in) and 181 lb gives BMI = 703 × 181 ÷ (70)² = 127,243 ÷ 4900 = 25.9 — same number, as expected. To move into the “healthy” band (BMI 24.9 or less), this person would need to weigh 78.9 kg (174 lb) or less, a 3 kg / 7 lb difference. To enter the “obese” class (BMI 30+), they would need to weigh 95 kg (210 lb). The arithmetic shows why BMI category transitions are sometimes a single meal — the band edges are flat numeric thresholds, not biological discontinuities.

A practical scenario: a clinician sees a 28-year-old patient at BMI 27 (officially “overweight”). The single number prompts a follow-up — waist circumference 80 cm (low), bioelectrical impedance shows 14% body fat, the patient lifts heavy weights five times a week. Conclusion: the BMI flag is a false positive driven by muscle mass. The same BMI 27 in a 55-year-old sedentary patient with waist circumference 110 cm and 32% body fat is a meaningfully different clinical picture. The number is identical; the underlying composition is not. This is the gap that BMI alone cannot bridge.

BMI across populations

The standard 18.5/25/30 cut-offs were calibrated against US and European population data. The WHO recommends Asian populations consider lower thresholds — overweight at BMI 23 and obese at BMI 27.5 — because cardiovascular and diabetes risk rises at lower BMI levels among East and South Asian groups. Pacific Islander and certain Polynesian populations show the opposite pattern, with healthy outcomes at higher BMI ranges. The implication: applying universal cut-offs to a diverse clinical population systematically over-flags some groups and under-flags others. Clinical guidelines in Japan, India, and Singapore use population-specific bands as a result.

When and why it matters

BMI matters as a five-second screening number for primary-care physicians, insurance underwriters, and epidemiologists studying population trends. It is the single most-collected health metric worldwide because it requires only a scale and a measuring tape, with no equipment, lab work, or training. It matters less for individual diagnosis: an NFL linebacker at 6′2″ / 250 lb registers as BMI 32.1 (“obese class I”) despite single-digit body fat. The opposite failure — “normal weight obesity”, where someone has BMI in the healthy range but excess visceral fat — is increasingly recognised by endocrinologists. The 2024 American Medical Association guidance explicitly cautioned against using BMI as the sole determinant of clinical decisions; insurance risk-scoring still uses it because nothing simpler has replaced it. Reference: CDC — About Adult BMI.

Why the square of height?Quetelet’s index was an empirical observation, not a physical derivation. If humans scaled isometrically (the same shape at every size), mass should grow with height cubed. Real humans don’t — taller people are proportionally thinner — and the squared term fits population statistics better than the cubed one does. Modern alternative indices like the Tri-Ponderal Mass Index (TMI = mass ÷ height³) match adolescent body fat slightly better and have been proposed as a replacement, but BMI’s century of clinical data makes it sticky.

Population vs individual — the persistent confusion:BMI was designed as a population-level statistic and its cut-offs (18.5, 25, 30) are calibrated against population averages of cardiometabolic risk. For any individual the metric is noisy — a 25-year-old male athlete with low body fat can land in the “overweight” band from muscle alone, while a sarcopenic 80-year-old can sit in the “healthy” band despite clinically meaningful lean-mass loss. Modern guidelines (WHO 2024, NIH) recommend pairing BMI with waist circumference and, where available, a body-composition measurement via DEXAor bioelectrical impedance. Reference: WHO — Obesity.