You have been handed a number that the scale, the app, or the doctor called your BMI, and you have been told it places you somewhere between 'normal' and 'obese'. The immediate question is whether that number is the truth about your body or a rough guess dressed in authority. The quick answer: BMI is a screening arithmetic for populations, not a diagnostic for individuals. The reason it feels wrong at your height is that its scaling rule is wrong. The core criticism is that BMI divides weight by height squared, but weight does not scale with height squared across all statures. It scales closer to height cubed in some groups, which is why the BMI prime and ponderal index alternatives exist at all.

To understand the fix, you need to see the original flaw. Adolphe Quetelet, a 19th-century mathematician, devised the Quetelet index as a statistical description of the 'average man', not as a health metric. He observed that in his sample, weight scaled with height squared, so he proposed a simple ratio of weight in kilograms divided by height in metres squared. A century later, physiologist Ancel Keys and his colleagues tested this index against alternatives in a comparative study published in 1972. They found it performed better than its rivals for population prediction. That is the moment the Quetelet index became the body mass index, and the moment the world inherited a scaling exponent derived not from physiology but from a snapshot of one man's data.

The Height-Scaling Problem

Quetelet's sample did not include everyone. When you scale weight by height squared, you assume that a taller person who is exactly twice as heavy as a shorter person at the same BMI has the same body shape. But an adult at 1.95 metres who weighs 95 kilograms has a BMI of 25.0. A 1.60-metre adult at 64 kilograms has the same 25.0. The taller person should, in many groups, carry more lean mass and less fat at the same weight-height ratio. This is the height-scaling problem: the squared term underestimates the weight of taller individuals and overestimates it for shorter ones. A short person can be misclassified as fatter than they are. A tall person can be told they are leaner. The mismatch matters. The WHO upper limit of normal, the 25 kg/m² threshold, was fixed using epidemiological cut-points from 1990s data, not from any metabolic law.

BMI Prime as a Translation Tool

This is where BMI prime enters. It is a dimensionless ratio that expresses how far someone is from the WHO upper limit. Take your BMI in kg/m² and divide it by 25. The result tells you what fraction of the way to that limit you stand. A BMI of 30 gives a BMI prime of 1.2. A BMI of 22 gives a BMI prime of 0.88. The threshold for overweight is a BMI prime of ≥1.0. For obesity class I it is ≥1.2, for class II it is ≥1.4, and for class III it is ≥1.6. The value here is not that it fixes the scaling error. It still uses the same squared height. The value is that it turns an awkward number like 31.7 into a proportion of a benchmark. It is a translation tool, not a new science.

Ponderal Index Versus BMI

The more interesting fix is the ponderal index, also called the Rohrer index. It takes the opposite approach: divide weight in kilograms by height in metres cubed. The normal range is often cited as 11-14 kg/m³ for adults. For neonates at term it is typically 2.2-3.0 g/cm³ × 100. The reasoning: if weight scales with height cubed, then the original Quetelet index is too lenient on tall people and too harsh on short ones. Keys' 1972 study, however, found that the ponderal index performed worse than the Quetelet index when predicting skinfold thickness across populations. That finding effectively killed the Rohrer index as a clinical contender for decades. It remains the standard measure of growth restriction in newborns, where the cubic scaling is more accurate.

If height cubed is too aggressive and height squared is too lenient, the truth lies somewhere in between. Allometric scaling suggests that weight scales with height raised to a power between 2 and 3, closer to 2.5 in many adult groups. That single observation explains why no simple power of height will ever satisfy everyone. The optimal exponent varies across ethnicities, ages, and levels of muscularity. It also explains why the search for a single perfect index is doomed: the relationship between height and weight is not a constant of nature but a statistical description of a specific group at a specific time.

Waist-Based Alternatives That Work

Recent proposals have stopped trying to fix the height exponent and instead added a second measurement. A Body Shape Index, known as ABSI, takes waist circumference and combines it with BMI and height to predict mortality risk. It was developed because waist circumference captures visceral adipose tissue, the fat stored inside the abdominal cavity around organs, which is independently associated with metabolic disease risk. The formula uses waist circumference divided by (BMI raised to two-thirds times height raised to one-half). It is a dimensionless ratio. ABSI has not been adopted by any major standard-setting body, and its absolute risk differences are small.

A more practical approach is the waist-to-height ratio. Divide waist circumference by height in the same units. The UK National Institute for Health and Care Excellence, in guideline CG189 updated in 2022, recommends a waist-to-height ratio of ≥0.5 as a central adiposity marker for adults with a BMI under 35 kg/m². Measure at the midpoint between the lowest rib and the iliac crest. This ratio is dimensionless, requires no exponent, and automatically adjusts for height in a way that no weight-based index ever will.

What the Evidence Actually Shows

The evidence for these alternatives is sobering. A prospective study of 894,576 adults in Western Europe and North America confirmed a J-shaped relationship between BMI and all-cause mortality. The same body of evidence shows that a given BMI predicts different fat percentages in different groups. The Global BMI Mortality Collaboration, drawing on 10,625,411 participants across 239 studies and 32 countries, found that the relationship between BMI and mortality is real but confounded by smoking, illness, and age.

Compare BMI to body fat percentage measured by DXA. The correlation is strong at population level but weak at individual level, especially in muscular individuals and older adults. This is the claimed-versus-real gap. Self-reported height and weight systematically bias BMI downward by 0.5 to 1.5 kg/m². Self-reported weight underestimates by 0.5 to 2.0 kg, with greater errors in women and those with higher BMI. A person can be overweight by BMI (25-29.9) and not overfat, or normal-weight by BMI and overfat. That state is called normal-weight obesity.

Which Measure to Use

No alternative has been adopted by a major standard-setting body to replace BMI. Each carries its own limitations. The WHO, CDC, and NICE still use BMI as the first-pass screening tool. It is free, it requires only height and weight, and it works at population scale. The ponderal index fails Keys' test. BMI prime is a useful way to talk about thresholds but does not fix the scaling. ABSI is too new and its clinical utility is unproven. The waist-to-height ratio of 0.5 is the single best addition to BMI that you can perform at home. It directly measures central fat distribution rather than inferring it from weight. Its measurement error is typically ±0.01 to 0.03, far smaller than the error of predicting fat from BMI.

If you are a healthcare professional explaining this to someone, stop presenting a BMI number as a verdict. Present it as one of several independent risk indicators. If you are looking at your own BMI, take out a tape measure. Measure your waist at the midpoint between the lowest rib and the iliac crest, and divide by your height. A result above 0.5 is a red flag regardless of what your BMI said. Below 0.5, you still need to know your blood pressure, fasting glucose, and lipid profile before you make any judgment. The BMI-to-mortality relationship is J-shaped: the nadir sits around 20 to 25 kg/m². The risks of the 'overweight' band are dramatically overestimated in observational data, driven by reverse causation and smoking.

The failure mode to watch for happens when you have only a scale and an app. The app will tell you that your BMI is 27.3 and your BMI prime is 1.09. If you are a muscular athlete, the app is wrong about you. The measurement error of BMI from self-reported data is real. The misclassification error of BMI as a body fat classifier is far worse. If you are 1.93 metres tall and weigh 100 kilograms, you are not 'overweight' in any clinically meaningful sense if most of that weight is lean mass. The only way to know is to measure your waist or your body fat percentage directly. If a DXA scan is unavailable, the tape measure around your waist is the most reliable proxy you have. It costs nothing and takes ten seconds.

What a Child's Centile Actually Tells You

If the BMI number belongs to a child, stop applying adult cut-offs. Paediatric BMI works differently: it uses centiles and z-scores derived from growth references, not absolute thresholds. A child at the 85th centile is not 'overweight' by the same definition as an adult at BMI 25. The WHO growth standards are prescriptive, meaning they describe how children should grow in optimal conditions. Many national charts are descriptive, meaning they show how children actually grew. The difference matters. A z-score of 2.0 on a descriptive chart is not the same as a z-score of 2.0 on a prescriptive chart. The failure mode here is comparing a child's BMI centile to an adult's BMI band as if they were the same scale.

If your child is above the 91st centile on the UK chart or the 95th on the CDC chart, you will be told they are 'overweight'. The evidence for intervention at that point is weak. The risk of creating an eating disorder is real. The better question is not whether the centile is high but whether it is crossing centiles over time, meaning the child is rapidly gaining weight relative to their height. That trajectory is a stronger warning than any single point. If the child is growing along their own centile, the number is likely a description of their body type, not a diagnosis of disease.

Why the Rohrer Index Never Caught On

The Rohrer index, despite its theoretical appeal, failed in the only test that mattered: Keys' comparative study of 1972. That study looked at thousands of men and compared how well various weight-height indices correlated with skinfold thickness, the best available proxy for body fat at the time. The Quetelet index won. It won because the cubic scaling of the Rohrer index over-corrected: it made tall people look too lean and short people look too fat. The practical consequence: if a clinician uses the Rohrer index on an adult today, they are using a tool that was ruled out by the same research that established BMI. The only place it survived is neonatology, where the newborn's body proportions are genuinely different, and where the index distinguishes symmetric from asymmetric growth restriction.

The lesson: an index that performs better in one group may perform worse in another. Keys himself warned against thinking of any index as a universal constant. The history of the ponderal index proves him right. If you are looking at a research paper that proposes a new weight-height index, check the population. If it was derived from a single ethnicity or a narrow age range, it will not generalise. The allometric scaling literature suggests the exponent changes with age, sex, and ethnicity. That is why the search for a single perfect exponent has failed for two centuries.

The body shape index takes a different route entirely by adding waist circumference to the height-weight relationship. Its developers showed that a higher ABSI, meaning more waist circumference for a given BMI and height, was associated with higher mortality in a large cohort. The effect sizes are real but modest. The index has not been picked up by any guideline body. A single tape measure around the waist gives you most of the information at zero computational cost. ABSI's complexity is a solution in search of a problem that a simple ratio already solves.

The waist-to-height ratio of 0.5 is the only alternative that has entered clinical guidelines, and even then only in the UK, and only as an addition to BMI, not a replacement. NICE CG189 recommends it for adults with BMI under 35. Above that threshold, the waist measurement adds little beyond the BMI's predictive power. The threshold is a single dimension: a 1.80-metre adult with a 90-centimetre waist is at the boundary. So is a 1.50-metre adult with a 75-centimetre waist. The ratio is dimensionless, so it needs no correction for sex or age. That is its main advantage over the WHO waist circumference thresholds, which are sex-specific and group-specific.

But even this measure has limits. The waist measurement protocol matters. The landmark (midpoint between lowest rib and iliac crest), the tension on the tape, and whether the patient is standing or supine all change the reading. The inter-rater reliability, meaning the standard deviation of repeated measures by different observers, is around 1 to 2 centimetres. That translates to a ratio error of about 0.01 to 0.02. Small enough for screening. Large enough to push someone right at the boundary from one side to the other. Measure yourself in the morning, without clothing, and take the average of three readings.

There is also the question of who the waist-to-height threshold does not fit. The 0.5 cut-off was derived largely from European and North American groups. Evidence suggests the optimal threshold is lower, around 0.45, for some Asian groups, and possibly higher for older adults. The WHO 2004 report on ethnic differences in the BMI-risk relationship documented that the same BMI carries a higher fat percentage and higher diabetes risk in South Asian and Chinese populations. The same is likely true for waist-to-height ratio. If you are using this number on someone of South Asian descent, a ratio of 0.48 may already be a risk marker even though it is below the NICE threshold.

The final alternative you will see mentioned is the body volume index, or some derivative of it, which uses 3D body scans to estimate body shape. These devices produce a wealth of data, but they are not a standardised measurement. No major body has endorsed them for clinical use. They are expensive, they are not portable, and the measurement error is not well characterised compared with the simple tape measure. If you have access to one, it is a novelty, not a diagnostic.

So the real answer to the question of what alternatives exist and whether they work is this. The alternatives that add a direct measure of central fat, like waist circumference or waist-to-height ratio, do add predictive value over BMI alone. They capture visceral adipose tissue, the fat that matters for metabolic disease. The alternatives that merely tinker with the height exponent, like the ponderal index, have been tested and failed. The alternatives that add computational complexity, like ABSI, have not yet shown they are worth the cost. Do not abandon BMI entirely. It is the only index with a century of population data behind it, and the risk thresholds are well calibrated for the groups they were derived from.

Your next step is the most practical thing you can do. Measure your waist today, at the midpoint between your lowest rib and your iliac crest. Divide that number by your height in the same units. If the answer is 0.5 or above, you have found a risk factor that your BMI alone was hiding. Discuss it with a clinician. If it is below 0.5, then your BMI is probably not the most important number on your health dashboard. Worry more about your blood pressure and fasting glucose. This single measurement takes less time than reading this page. It costs nothing. It is the one adjustment to the BMI framework that clinical guidelines have actually adopted.

Do not spend money on a smart scale that promises to calculate your BMI prime or your ponderal index. The arithmetic is trivial and you can do it in your head. Do not switch to an app that tells you your 'metabolic age' or your 'body shape index'. These are not standardised measurements and the reference data are shaky. The only tool worth carrying is a fabric tape measure. The only number you need to remember is 0.5. Everything else is a statistical description of a group that is not you.

If you are a professional explaining this to a client or a patient, the message is the same. BMI is a screening tool. Waist-to-height ratio is a risk marker. Neither is a diagnosis. The J-shaped curve between BMI and mortality has a nadir around 22. That does not mean that everyone below 22 is healthy or that everyone above 25 is sick. The only way to know an individual's risk is to measure their blood pressure, their fasting glucose, their lipids, and their waist, and to ask about their family history and their smoking status. No index, however sophisticated, replaces that conversation.

The history of this subject is a history of failed attempts to find a single number that captures body composition. Quetelet found a statistical regularity. Keys validated it. The WHO standardised it. A hundred researchers have since tried to improve it, with limited success. The reason: the relationship between height and weight is not a biological law. It is a statistical description of a particular group at a particular time. Accept that, and you stop looking for a perfect index. You start using the imperfect ones wisely.

The one failure that will set you back is treating any of these numbers as fate. A BMI of 31 is not a death sentence. A waist-to-height ratio of 0.51 is not a one-way ticket to metabolic disease. These numbers are risk factors. They describe the odds of a group, not the destiny of an individual. Use them as signals that something might be worth checking, and then check it with a more direct measure. That is the entire point of screening.

BMI Prime, the Ponderal Index, and Other Variants That Address BMI's Scaling Problem