Is BMI a good measure of health? A BMI of 30 misses more than half of excess fat

Not on its own. Body mass index is cheap, reproducible and useful for comparing populations, but as a test of whether an individual carries harmful excess fat it is weak — it misses a large share of the people who have it, and in 2025 an international commission recommended it no longer be used by itself to diagnose obesity in a person [s1] [s2].

What BMI actually measures

BMI is weight divided by height squared. It says nothing directly about how much of that weight is fat rather than muscle, or where the fat sits. The clearest demonstration of the gap comes from a study of 13,601 US adults in the Third National Health and Nutrition Examination Survey, which compared BMI against body-fat percentage measured by bioelectrical impedance, using the World Health Organization thresholds for obesity of over 25% body fat in men and over 35% in women [s2].

By BMI (≥30 kg/m²), obesity was present in 19.1% of men and 24.7% of women; by measured body fat, it was present in 43.9% of men and 52.3% of women [s2]. A BMI of 30 or above was highly specific — 95% in men and 99% in women — meaning few lean people are wrongly flagged [s2]. But its sensitivity was poor: 36% in men and 49% in women [s2]. In the study's own words, the cutoff "has good specificity but misses more than half of people with excess fat" [s2]. The failure was worst in the intermediate BMI range of 25 to 29.9, where the measure could not separate fat mass from lean mass in either sex, and it worsened with age [s2].

The 2025 rethink

In January 2025 a Lancet Diabetes & Endocrinology commission of 58 experts across multiple specialties and countries — including people with lived experience of obesity — published new diagnostic criteria [s1]. Its core judgement was that current BMI-based measures can both underestimate and overestimate adiposity and provide inadequate information about health at the individual level [s1].

The commission recommended that BMI be used only as a surrogate measure of risk at the population level, for epidemiology or screening, rather than as an individual measure of health [s1]. Excess adiposity, it said, should be confirmed either by direct measurement of body fat or by at least one anthropometric criterion — waist circumference, waist-to-hip ratio, or waist-to-height ratio — in addition to BMI [s1]. There is one pragmatic exception: at a BMI above 40 kg/m², excess fat can be assumed and no further confirmation is required [s1].

It went further and split the diagnosis in two. "Clinical obesity" it defines as a chronic illness in which excess fat is already impairing the function of organs or tissues, or substantially limiting daily activities; "preclinical obesity" is excess fat with organs still working normally but a raised future risk [s1]. The recommendations were agreed with 90–100% consensus among the commissioners and endorsed by 76 organisations worldwide [s1].

The better numbers to know

For most people the practical upgrade is a tape measure. Waist circumference and waist-to-height ratio capture abdominal fat, the depot most tied to metabolic disease, which BMI ignores entirely [s1]. This is why two people with an identical BMI can carry very different risk, and why a muscular person can be labelled obese by BMI while carrying little fat — the specificity works in that direction too [s2].

None of this makes BMI useless. It remains a reasonable population screen, it is quick and free, and above 40 kg/m² it is informative on its own [s1]. The mistake is treating a single number, built for describing groups, as a verdict on an individual body. The evidence says pair it with a measure of where the fat is, and, if the question is disease rather than weight, ask whether any organ is actually affected [s1] [s2].

A last nuance worth holding onto: BMI's errors run in both directions. It over-diagnoses in the muscular and, far more often, under-diagnoses in people who carry fat at a "normal" weight — the group sometimes missed entirely because the scale looks reassuring [s2]. Neither error is a reason to ignore weight; both are reasons to measure the thing you actually care about, which is fat and what it is doing, not the ratio of mass to height.

This article is informational and is not medical advice.

Sources

• Definition and diagnostic criteria of clinical obesity — The Lancet Diabetes & Endocrinology, 2025-01-14
• Accuracy of body mass index in diagnosing obesity in the adult general population — International Journal of Obesity, 2008-02-19