Four formulas, one shape
Every formula on this page has the same structure: a base weight at five feet, plus a fixed increment for each inch above that. Only the constants differ.
| Formula | Male | Female | Origin |
|---|---|---|---|
| Hamwi, 1964 | 48.0 kg + 2.7 kg/in | 45.5 kg + 2.2 kg/in | Bedside rule for diabetic diet planning |
| Devine, 1974 | 50.0 kg + 2.3 kg/in | 45.5 kg + 2.3 kg/in | Scaling aminoglycoside doses to lean mass |
| Robinson, 1983 | 52.0 kg + 1.9 kg/in | 49.0 kg + 1.7 kg/in | Refit of Devine to insurance-table data |
| Miller, 1983 | 56.2 kg + 1.41 kg/in | 53.1 kg + 1.36 kg/in | Another refit, flatter slope |
Because the slopes differ, the four agree most closely near five feet and diverge as height increases. At six foot two the spread between the highest and lowest is several kilograms for the same person. There is no principled way to choose between them, which is a reasonable clue about how much weight any of them should carry.
Devine was a dosing equation
This is the part that usually surprises people. B. J. Devine published the formula in 1974 in a pharmacy journal, addressing a specific clinical problem: some drugs distribute poorly into fat tissue, so dosing them on total body weight overdoses heavy patients. Devine needed a quick estimate of lean mass from height, and the constants were chosen to make the arithmetic convenient rather than fitted to a dataset. The formula went on to be validated for that purpose and is still used for it.
What it was never validated for is telling a person what they should weigh. There is no cohort behind it, no outcome it was optimised against, no morbidity curve it sits on. Its appearance on consumer calculators is a case of a clinical tool escaping its context and acquiring an authority it was never given. Hamwi has a similar story — a rule of thumb from a diabetes clinic, published as a chapter in a 1964 handbook.
What the BMI span is doing differently
The 18.5 to 25 range shown at the top of the output has a different pedigree. It comes from epidemiological work relating BMI to mortality and disease incidence across large populations, which is a weak instrument for an individual but at least points at health outcomes. It is a range rather than a point, which is the honest shape for this kind of estimate, and at a given height it typically spans thirty to forty pounds.
The classic formulas all land somewhere inside that range, usually in the lower half of it. That is not evidence of their accuracy; it mostly reflects that both were derived from populations of similar build. What matters more is the difference in form. A single number invites the question "how far am I from it", which is the wrong question. A thirty-pound band invites "am I inside it", which is at least answerable and, for most people, already answered yes.
Percent of ideal body weight
If you enter a current weight the output shows the ratio to the Devine figure as a percentage, because that number appears in clinical notes and nutrition assessments and people encounter it on paperwork. Traditionally, under 90 percent gets flagged as underweight, 90 to 110 as within range, and progressively higher bands above that. It is used most often in nutritional screening of hospitalised patients, where a rapid change in the ratio is the signal being watched, rather than the level itself.
Outside that context the figure is not doing much work that BMI does not do better, and it carries the added problem of being anchored to a dosing formula. It is shown for recognition rather than recommendation. If you have seen it on a discharge summary and wanted to know what it meant, that is the answer; it is not a target to move towards.
Questions people ask
Which of the four formulas is the right one?
None of them is right in the sense the question implies, because none was validated against health outcomes. Devine is the one to use if you specifically need the number a clinical pharmacy calculation would use, since that is what it was built for. For thinking about your own weight, the BMI-derived range shown above the formulas is a better reference, and it is a range rather than a point for a reason. The value in showing all four is that their disagreement makes the precision of any single one look as questionable as it is.
The formulas give a lower number than my BMI range. Which do I follow?
That is the expected pattern — the classic formulas mostly sit in the lower part of the 18.5 to 25 band, because they were fitted to mid-twentieth-century insurance and clinical data. It does not mean the lower number is a target. A weight anywhere inside the BMI span is, on population data, associated with similar risk, and where an individual sits within it depends on build, muscle mass and history rather than on which formula they read. Nothing on this page is a recommendation to weigh a particular amount.
Do these work for athletes or very muscular people?
Poorly. Every formula here is height in, weight out, with no term for body composition at all. A rugby forward or a competitive lifter will exceed all four figures substantially while carrying a low body fat percentage, and the formulas have no way to notice. This is the same blind spot BMI has, for the same reason. Where composition is the actual question, a tape-based or laboratory estimate of body fat is the relevant measurement, and weight is only meaningful next to it.
Why are the male and female constants different?
Because the source data was split by sex and the fitted constants came out different, mostly reflecting average differences in lean mass and frame at a given height. The size of the difference varies by formula, which is itself telling: Devine uses the same per-inch slope for both and differs only in the base, while Robinson and Miller differ in both terms. There is no biological principle producing those specific numbers. They are the coefficients that fell out of four different datasets, which is why they do not agree with one another.