What BMI actually is
Body Mass Index was invented in 1832 by Adolphe Quetelet, a Belgian statistician, mathematician, and astronomer — not a physician. His goal was to describe the statistical distribution of body weight in large populations, not to diagnose individual health. The formula — weight in kilograms divided by height in metres squared — was a convenient approximation because it minimised the correlation between weight and height in a population sample. It was never designed as a clinical tool.
The World Health Organisation formally adopted BMI as an international obesity classification standard in 1995, with cutoffs set at 18.5, 25, 30, and 40. Those numbers were chosen by expert consensus, not derived from sharp inflection points in mortality data. Since 1995, BMI has become the dominant screening metric in primary care — largely because it requires nothing more than a scale and a tape measure.
BMI categories explained
The WHO defines five categories using the same cutoffs globally. However, the Asia-Pacific region uses lower thresholds — 23 for overweight and 27.5 for obesity — because research consistently shows that Asian populations develop metabolic risk at lower BMI values than European populations, likely due to differences in body fat distribution and muscle mass.
The 5 cases where BMI lies
BMI is a population average masquerading as an individual measurement. Here are the five populations for whom it is most systematically wrong:
- Athletes and muscular people. Skeletal muscle is denser than fat. A 90 kg sprinter with 10% body fat and a 90 kg sedentary office worker with 32% body fat have identical BMIs. The WHO "Obese Class I" label — applied to both — is medically meaningful for one and clinically irrelevant for the other. NFL linemen, powerlifters, and competitive cyclists routinely score 28–35 despite elite cardiorespiratory fitness and minimal visceral fat.
- Elderly adults. After age 65, muscle mass and bone mineral density decline significantly. This can produce a "normal" BMI — say 22 — while masking sarcopenic obesity: a high body fat percentage with loss of functional muscle. Worse, studies in elderly populations consistently show a survival advantage in the BMI 25–30 range (the "obesity paradox"), which directly contradicts standard BMI-based risk messaging.
- Pregnant women. BMI becomes meaningless during pregnancy. Weight gain is expected, necessary, and healthy. Pre-pregnancy BMI matters for gestational weight gain targets, but during pregnancy BMI-derived risk stratification is not clinically applicable.
- Pediatric populations. BMI cutoffs for adults do not apply to children and adolescents. For those under 18, physicians use BMI-for-age percentile charts that account for normal growth patterns across sex and developmental stage.
- Certain ethnicities. South Asian, East Asian, and South-East Asian populations develop insulin resistance and cardiovascular risk at lower BMI values than European populations, prompting the WHO to issue Asia-Pacific cutoffs. Conversely, Black populations may have higher lean mass at equivalent BMI, meaning standard cutoffs overclassify risk. A single global cutoff applied to a globally diverse population is an oversimplification with real clinical consequences.
Better metrics than BMI alone
These four measures, used together, give a substantially more complete picture of metabolic risk than BMI can provide on its own:
Waist circumference directly correlates with visceral fat — the metabolically active fat surrounding abdominal organs. Visceral fat drives insulin resistance, inflammation, and cardiovascular risk far more than subcutaneous fat. Measure at the midpoint between the lowest rib and the iliac crest, relaxed, after exhaling.
Divides waist circumference by hip circumference. A high ratio indicates an apple-shaped fat distribution pattern, which carries higher cardiovascular risk than a pear-shaped pattern. Better than waist circumference alone in tall individuals.
Remarkably robust across ethnicities, age groups, and sexes. The mnemonic: "keep your waist to less than half your height." A 170 cm person should target a waist below 85 cm. This metric outperforms BMI in predicting cardiometabolic risk in multiple large meta-analyses.
The only metric that directly measures what you actually care about. Gold standard methods: DEXA scanning (dual-energy X-ray absorptiometry), hydrostatic weighing, and BodPod (air displacement plethysmography). Consumer bioelectrical impedance scales are less accurate but directionally useful for tracking trends over time.
When BMI is actually useful
None of this means BMI is worthless. It is genuinely valuable in two contexts:
Population epidemiology. BMI is cheap, reproducible, non-invasive, and correlates well with health outcomes at the population level. It allows researchers to compare weight distributions across countries and decades. For this purpose — studying millions of people — it performs adequately.
Baseline clinical screening. In a primary care setting where a physician has five minutes, BMI is a reasonable first filter. A BMI of 17.2 or 42 should prompt further investigation. The problem is when it becomes the conclusion rather than the starting point — when a BMI of 26 ends the conversation rather than beginning a more thorough metabolic assessment.
BMI and diabetes risk
The American Diabetes Association's Standards of Medical Care 2024 recommends testing for type 2 diabetes in adults aged 35–70 who are overweight or obese (BMI ≥ 25), and in adults of any age with additional risk factors including a first-degree relative with diabetes, a history of gestational diabetes, or signs of insulin resistance. For Asian Americans, testing is recommended at BMI ≥ 23, reflecting the higher diabetes risk at lower BMI values documented in this population.
This is one of the most legitimate clinical uses of BMI: not to diagnose diabetes or even to quantify metabolic risk precisely, but as an inexpensive threshold that triggers appropriate diagnostic testing. The actual diabetes diagnosis and risk stratification rely on fasting plasma glucose, HbA1c, and the oral glucose tolerance test — not BMI itself.
If you are managing calorie intake as part of a diabetes prevention or management strategy, the CalEye for diabetes page covers how accurate calorie tracking supports glycaemic control without the administrative friction of manual food logging.
Related tools and reading
Frequently asked questions
Is BMI accurate for athletes and muscular people?
No. BMI cannot distinguish between muscle mass and fat mass. A competitive rugby player or powerlifter will often score in the "Overweight" or "Obese" range despite having very low body fat. The formula treats two kilograms of muscle identically to two kilograms of adipose tissue, which is physiologically nonsensical. Athletes should use DEXA, BodPod, or hydrostatic weighing to assess actual body composition.
Does BMI apply accurately to elderly people?
Not reliably. Older adults lose muscle mass (sarcopenia) and bone density with age, which can produce a "normal" BMI while hiding dangerous levels of visceral fat and functional weakness. Studies show that elderly individuals at a BMI of 23–27 often have better outcomes than those at 18.5–22 — the "obesity paradox." Waist circumference and functional strength tests are more informative in this group.
Why is BMI an imperfect measure of health?
BMI was invented in 1832 as a population-level statistical tool, not a clinical diagnostic. It measures weight-to-height squared — ignoring body composition, fat distribution, age, sex, and ethnicity. Two people with identical BMIs can have dramatically different metabolic risk profiles depending on where fat is stored and how much of their "weight" is muscle.
What metrics do doctors actually use instead of BMI alone?
Waist circumference (elevated risk: ≥ 94 cm in men, ≥ 80 cm in women), waist-to-hip ratio, waist-to-height ratio (target below 0.5), and direct body fat percentage measurement via DEXA, bioelectrical impedance, or hydrostatic weighing. These metrics better predict visceral fat — the metabolically active fat linked to insulin resistance, cardiovascular disease, and type 2 diabetes.
Is BMI 25 actually overweight — should you worry?
Not necessarily. A BMI of 25–26 with a healthy waist circumference, good cardiorespiratory fitness, normal blood glucose, and no hypertension carries very little elevated risk. The cutoff of 25 was set by committee consensus, not by a sharp inflection in population health outcomes. What matters is your metabolic health markers — blood pressure, fasting glucose, lipid panel — not a single anthropometric ratio.