How Body Mass Index Misreads Very Short and Very Tall Bodies

Body mass index divides weight by height squared. Bodies scale in three dimensions, so the index overestimates adiposity in very short people and underestimates it in very tall people. That distortion is not a flaw in your scale. It is a mathematical consequence of using a two-dimensional ratio on a three-dimensional object.

The Quetelet index, the 1830s forerunner of BMI, was never designed to assess individual body composition. It described the statistical properties of the average man. When Ancel Keys examined the index in 1972, he found it was selected for its association with population adiposity, not for accuracy at height extremes. The choice of height squared over height cubed was pragmatic, not physiological.

A very short adult with normal body fat can land in the overweight band. A very tall adult with excess body fat can land in the normal band. The error is systematic, not random, and it grows as you move away from the average height around which the index was calibrated.

The BMI formula behaves correctly when body proportions are average. Bodies do not scale isometrically with height. A person 20 percent taller than average will not have a trunk 20 percent larger; they will have longer limbs. Lean mass scales with height cubed, while the Quetelet index divides by height squared. That mismatch means the index is contaminated by stature.

The Rohrer index uses height cubed: weight in kilograms divided by height in metres cubed. It removes the height relationship that plagues BMI. The Rohrer index was used in paediatrics before BMI became dominant, and it is still the more accurate index for children whose bodies are still scaling in three dimensions. For adults, the Rohrer index produces a flatter relationship with height. It does not penalise short people or reward tall ones.

Ancel Keys's 1972 paper documented this. He showed that the association between BMI and height changed sign depending on the population, a sign that the index was not measuring what it claimed to measure. The allometric scaling problem, how mass relates to length, was known before the WHO adopted the index as a screening tool.

Short Adults and the Overweight Misclassification

The distortion is not trivial. At low statures, the overestimation of body fat by BMI becomes large enough to shift a person into a higher WHO weight category. A woman well below average height with body fat in the healthy range by DXA can land above the WHO 1995 overweight cut-off published in Technical Report Series 854. A man of short stature with normal body fat can be classified as pre-obese.

Tall Adults and the Normal-Weight Blind Spot

The reverse applies at the tall end. At high statures, BMI underestimates adiposity. A very tall man with body fat above the obesity threshold of 25% for men set by the American College of Sports Medicine can register a BMI that places him only in the overweight band. The WHO 2000 reaffirmation in Obesity: Preventing and Managing the Global Epidemic (TRS 894) did not address this scaling problem, because the cut-offs were derived from epidemiological mortality data, not from individual body composition.

The Benn Index and the Case Against a Fixed Exponent

The Benn index, proposed in 1971 by statistician A. G. Benn, replaces the fixed power of 2 with a population-specific power that produces an index uncorrelated with height. In short populations, that power is often closer to 1.6. In tall populations, it can be as high as 2.5. The fact that the correct power varies by population is the strongest evidence that the fixed power of 2 is wrong for individuals at the height extremes.

Index	Formula	Height Exponent	Correlation With Height	Best Use
BMI (Quetelet index)	Weight / Height²	2.0	Positive in short populations, negative in tall	Population screening, average heights
Rohrer index	Weight / Height³	3.0	Near zero	Paediatric growth, height-extreme adults
Benn index	Weight / Height^p	1.6–2.5 (population-specific)	Zero by design	Research populations with known allometry

The Rohrer index is the simplest alternative that addresses the height scaling problem. By dividing by height cubed, it assumes mass scales with volume, which is geometrically sounder than height squared. In a sample of adults where BMI correlates positively with height, meaning tall people have higher BMIs purely because they are tall, the Rohrer index removes that relationship.

The Rohrer index was the standard paediatric index in continental Europe before BMI replaced it. In children aged 2 to 10, whose growth is still dominated by linear expansion, height cubed is a better predictor of body mass than height squared. The WHO international growth standards for under-5s use a different approach, weight-for-length rather than weight-for-height, precisely because the squared power fails at small statures.

For very short adults, switching from BMI to the Rohrer index reclassifies a meaningful fraction from overweight to normal. For very tall adults, it reclassifies some from normal to overweight. The clinical effect is to bring the index into line with body fat percentage measured by DXA, the reference method for body composition in research.

NHANES and the Short-Woman Problem

The misclassification at height extremes is not a niche theoretical point. In the NHANES dataset, the fraction of short women misclassified as overweight by BMI while having normal body fat by DXA is large enough to affect obesity prevalence statistics. The Diverse Populations Collaborative Group, in a 2005 paper, showed that the height-BMI relationship varies by ethnicity. The misclassification is not uniform across populations.

Asian Populations and the Double Burden

Asian populations, who already have lower BMI thresholds for increased risk from the WHO 2004 consultation, face a double problem: their average stature is shorter, and the BMI scaling error pushes more individuals into a higher classification. The WHO Asian-specific thresholds, normal range 18.5-22.9 kg/m², increased risk at ≥23.0, high risk at ≥27.5, were issued in the 2004 Lancet publication after the expert consultation. Those cut-offs partially compensate for the height scaling problem because they shift the classification downward, but they do not fix the fundamental power error for individuals at the extremes.

Frame Size: The Forgotten Variable

Skeletal frame size compounds the problem. The Frisancho elbow breadth measurement is one method to classify frame size as small, medium, or large. Two people of the same height and weight but different frame sizes will have different body fat proportions, yet BMI treats them identically. The Metropolitan Life 1983 tables used height and frame size together; BMI replaced that approach but lost the frame size adjustment.

Allometric scaling describes how body mass changes with body length across species and across individuals within a species. In humans, the power is not a universal constant. Ancel Keys's 1972 finding was that the relationship between the Quetelet index and height varied so widely across populations that the index could not be a valid measure of adiposity independent of stature.

BMI prime, the ratio of a person's BMI to the WHO normal upper limit of 25, inherits the same height relationship. Multiplying the index by a constant does not change its shape. The only way to remove the height relationship is to change the power, which is what the Rohrer index and the Benn index do.

In research settings, the four-compartment model is the gold standard for body composition. It combines body density from underwater weighing, total body water from deuterium dilution, and bone mineral content from DXA. That model is impractical for screening, but it reveals the error in BMI. A short person with a BMI of 27 and a tall person with a BMI of 27 can have very different body fat proportions, and the difference is driven by the height scaling problem.

If you are very short or very tall, stop relying on BMI for individual assessment. Two direct measurements bypass the scaling problem: body fat percentage and waist-to-height ratio.

Body fat percentage is the proportion of total mass that is adipose tissue. For men, the ACSM consensus threshold for obesity is greater than 25% body fat. For women, the threshold is greater than 35%, though this figure depends on the measurement method and no single survey provides a definitive population number. DXA, BIA, and skinfold calipers give different results, but none of them depend on height. A short person with 22% body fat is not overfat, regardless of what BMI says.

If you are below 152 cm or above 183 cm, ignore your BMI number for personal health assessment. Book a waist-to-height ratio measurement with a trained observer: a clinician, a physiotherapist, or a registered dietitian. If that is not available, book a DXA scan for body fat percentage. The BMI number is a population screening tool. At the extremes of height, it misleads.

The most common failure is the short person who believes their BMI is telling them they are overweight when their body fat percentage is normal. They restrict calories unnecessarily, lose lean mass, and end up with a worse metabolic profile. The tall person who believes their normal BMI means they have no excess body fat delays addressing an actual problem. Switch measurement systems: use waist-to-height ratio and body fat percentage, not BMI, to understand your body.