Body Frame Size, and What Ideal-Weight Formulas Assume

Why frame size was invented

Height fixes very little about a skeleton. Two people who both measure 178 cm can differ measurably in the width of the pelvis, the breadth of the shoulders, the thickness of the long bones and therefore the mass of bone they carry. Insurance height-weight tables in the twentieth century ran straight into this: their weight bands were narrow, and applying one band to everyone of a given height produced obvious mismatches. Frame size was the patch. Split each height into three categories of skeletal build, publish a different weight band for each, and the tables stop insulting people at the edges of the distribution.

That origin is worth holding on to, because it explains what frame size is good for and what it is not. It was invented to make a weight table behave better, not to describe anything about a person's health. It is a correction term. And like most correction terms, it turns out to be far cruder than the confidence with which it is usually reported.

The two published methods, and their actual cut-offs

The wrist method uses a ratio that nutritional assessment literature calls r: height divided by wrist circumference, both measured in the same unit. The wrist is chosen because it carries almost no muscle and very little fat, so its circumference is close to a direct read of bone. For men, r above 10.4 indicates a small frame, 9.6 to 10.4 a medium one, and below 9.6 a large one. For women the bands shift: above 11.0 small, 10.1 to 11.0 medium, below 10.1 large. Made concrete, a man of 178 cm falls in the medium band with a wrist between 17.1 and 18.5 cm; a woman of 165 cm falls in hers between 15.0 and 16.3 cm.

The elbow method measures skeletal width directly: with the arm bent forward at a right angle and the fingers pointing up, the distance across the two prominent bones of the elbow. The reference bands came out with the 1983 Metropolitan Life tables and were later grounded in a much larger multiracial dataset by Frisancho, working from over twenty thousand NHANES participants. The bands themselves are startlingly narrow. For a man of 5'8" to 5'11" the medium range is 2¾ to 3 inches, which is 6.99 to 7.62 cm; for a woman of 5'4" to 5'7" it is 2¼ to 2½ inches, or 5.71 to 6.35 cm. In both cases the entire medium category is a strip 0.64 cm wide.

That last figure is the honest verdict on the whole exercise. Six millimetres of measurement error — the width of a tape measure held slightly wrong, or an elbow caliper set a degree off — moves a person from one frame category to the next. The categories are real in the sense that they were derived from real anthropometric data. They are not precise enough for anyone to build a self-image on.

The ideal-weight formulas were built for dosing

Here is the fact that reframes everything. Devine's equation — the most cited ideal-weight formula in the world — first appeared in 1974 in a note about calculating an antibiotic dose. Its purpose was to give clinicians a lean-mass proxy for drugs that do not distribute into fat. Robinson and colleagues published theirs in 1983 in a hospital pharmacy journal, in a paper whose title says outright that it is for drug dosage calculations. Hamwi's, from 1964, came out of dietetic practice as a bedside rule of thumb. None of them was derived from mortality data, from morbidity data, or from any study of what weight is associated with living longer.

All four share a structure: a base weight at five feet, plus a fixed amount for every inch above that. Hamwi for men is 48 kg plus 2.7 kg per inch; Devine is 50 plus 2.3; Robinson is 52 plus 1.9; Miller is 56.2 plus 1.41. For women the pairs are 45.5 plus 2.2, 45.5 plus 2.3, 49 plus 1.7 and 53.1 plus 1.36. Notice what varies: not just the intercept but the slope, and the slopes differ by nearly a factor of two between Hamwi and Miller. That is why they cannot possibly agree across a range of heights. They were each fitted to a different dataset with different purposes, and none was ever meant to be compared with the others.

How far apart they actually get

Run all four across a range of heights and a clear pattern appears. For men at 5'6" (168 cm) the four answers are 64.2, 63.8, 63.4 and 64.7 kg — a spread of 1.3 kg, essentially agreement. Push to 6'2" (188 cm) and they give 85.8, 82.2, 78.6 and 75.9, a spread of 9.9 kg. At 6'4" (193 cm) the gap is 12.4 kg, or 27 lb, between Hamwi at the top and Miller at the bottom. Go the other way, to 5'2" (157 cm), and the order inverts: Miller is now the highest at 59.0 and Hamwi the lowest at 53.4.

The women's set behaves the same way and crosses a little higher. At 5'8" (173 cm) the four give 63.1, 63.9, 62.6 and 64.0 kg, a spread of 1.4 kg. At 4'10" (147 cm) they give 41.1, 40.9, 45.6 and 50.4 — a spread of 9.5 kg, or 21 lb, on the same short woman. Miller's formula, with its high base and shallow slope, says nearly ten kilograms more than Hamwi's at that height and nearly three kilograms less at 6'0". They are simply four lines with different slopes, crossing near the middle of the range each was fitted on and diverging steadily from there in both directions.

What BMI adds, and what it hides

It is worth putting the four formulas next to BMI, because the comparison is revealing in both directions. Convert each ideal weight into the BMI it implies and the differences stand out sharply. For a man of 163 cm the four sit at BMI 22.3, 22.4, 22.6 and 23.4 — all within a couple of points. For a man of 193 cm they sit at 24.5, 23.3, 22.1 and 21.1, a range of over three BMI points from the same four equations. Miller's shallow slope means it implies a lower and lower BMI as height rises, which is a strong claim about body proportions that nobody making the formula set out to assert.

BMI's advantage over all of them is that it was studied against health outcomes in large populations, so its bands mean something at population scale that an insurance-era regression does not. Its limitation is equally well documented and equally important: it cannot distinguish muscle from fat, it says nothing about where fat sits, and its cut-offs perform differently across ancestries — which is why several countries publish adjusted thresholds. Frame size, the four ideal-weight formulas and BMI are all population instruments. Each is useful for the job it was built for. None of them is a description of an individual person, and none of them is a target.

Why they are still used, and why not as a target

The formulas survive because the job they were built for still exists. A drug that distributes into lean tissue and water rather than into fat needs a lean-mass denominator, and ideal body weight is the cheapest available estimate of one — it takes a height and a sex and nothing else. That is exactly the use described in the weight-based dosing article, where the same 178 cm, 120 kg patient gets doses 64% apart depending on which weight is chosen. The formulas are load-bearing in clinical pharmacy, and their disagreement matters there for reasons that have nothing to do with how anyone looks.

Reading them as a personal weight target is a category error, and worth naming as such. A formula fitted so that an antibiotic dose lands in a therapeutic window is not, and was never claimed to be, a statement about what weight is healthy for a human being. It has no term for muscle mass, for age, for activity, for metabolic health, for anything measured in a laboratory. It cannot distinguish an athlete from a sedentary person of identical height. And when four such formulas disagree by 27 pounds on the same tall man, treating any one of their outputs as a number to reach means picking, arbitrarily, between four historical accidents.

Frequently asked questions

How do I measure my frame size?

The wrist method needs two numbers in the same unit: your height and the circumference of your wrist just below the bony prominence, on the side you write with. Divide the first by the second. For men, a result above 10.4 is a small frame, 9.6 to 10.4 medium and below 9.6 large; for women the thresholds are 11.0 and 10.1. Be aware that the categories are so narrow that a few millimetres of tape-measure error can change the answer, so treat the result as approximate.

Does having a large frame mean I should weigh more?

Frame size was designed to widen or narrow the weight band that an insurance table assigned to a given height, so within that historical system a larger frame corresponded to a higher band. But that system was not a health target and this article is not offering one. What you should weigh is not a question a wrist measurement can answer, and it is not a question this page will answer. If it matters to you, a doctor or a registered dietitian who can actually see you is the right place to take it.

Which ideal-weight formula is the most accurate?

The question has no answer, because none of them is estimating a measurable quantity. They are not approximations of a true ideal weight that someone could go and measure; they are four regressions fitted for different purposes on different data. Devine dominates in clinical pharmacy because it was there first and dosing conventions were built on top of it, not because it was shown to be better. Their disagreement — up to 12.4 kg at 193 cm — is the honest answer to how much precision any of them has.

Why do the formulas use inches when I measure in centimetres?

All four were written in the United States between 1964 and 1983, for a clinical world that recorded height in feet and inches. None has been re-derived in metric, so a metric implementation converts first and inherits the rounding. The effect is small but non-zero: 178 cm is 70.08 inches, so rounding to a flat 70 costs about 0.2 kg on Devine's answer. It is one more reason two calculators can differ slightly on the same person without either being wrong.

Is BMI better than these formulas?

For anything to do with health, yes, in one specific respect: BMI bands were studied against outcomes in large populations, whereas the ideal-weight formulas were fitted for clinical arithmetic and never tested against mortality or morbidity at all. But BMI has its own well-documented limits — it cannot separate muscle from fat, it says nothing about fat distribution, and its cut-offs behave differently across ancestries, which is why some countries publish adjusted thresholds. Both are population tools. Neither describes an individual.