The arithmetic, and where the 703 comes from
Body mass index is defined in metric units: mass in kilograms divided by height in metres squared. Working in feet and pounds you either convert, which is what this page does, or you use the shortcut version — weight in pounds times 703, divided by height in inches squared. The 703 is not a physical constant, it is simply the conversion factors folded together, and it is the reason the imperial form looks arbitrary.
Take 5 feet 10 inches and 170 pounds. That is 70 inches and, converting exactly at 2.54 centimetres to the inch and 0.45359237 kilograms to the pound, 1.778 metres and 77.11 kilograms. Divide 77.11 by 1.778 squared, which is 3.1613, and you get 24.4. The shortcut agrees: 170 times 703 is 119,510, divided by 4,900 is 24.4.
Why the boundaries move depending on who printed the chart
The 18.5, 25 and 30 lines are a WHO convention that most English-language sources follow. They are not the only set in use. In 2004 the WHO published an expert consultation noting that in many Asian populations the health risks associated with a given BMI appear at lower values, and set out additional public-health action points at 23 and 27.5. Several national bodies adopted those, and clinical guidance in a number of countries uses 23 as an overweight threshold rather than 25. The same reading of 24 is inside the range on one chart and above it on another.
Neither set is more correct than the other in the abstract. They are drawn from different population data with different distributions of body composition, and both are thresholds chosen to be useful for looking at groups. Which one an individual should be read against is a clinical question, not a formula question, which is why both are shown here rather than one being picked for you.
What the index cannot see
BMI has one input for body composition, which is total weight, so it treats a kilogram of muscle and a kilogram of fat as the same thing. The consequences run in both directions.
| Situation | What BMI does |
|---|---|
| Substantial muscle mass from training | Reads high; the weight is there but very little of it is fat |
| Little muscle, weight in the middle of the range | Reads normal while body fat percentage may be well above average |
| Fat carried around the abdomen | Invisible to the index, though it is the distribution most associated with metabolic risk |
| Older adults losing muscle | The index can fall or hold steady while composition changes considerably |
| Very tall or very short adults | The squared-height denominator fits the middle of the range better than the ends |
Waist circumference is the cheap addition that fixes some of this, because it responds to where the weight is rather than only how much of it there is. A tape measure at the navel, taken standing and after a normal breath out, gives a second number that is independent of the first. Neither one diagnoses anything on its own.
What it was built for
The ratio comes from Adolphe Quetelet, a Belgian statistician working in the 1830s who was describing the distribution of measurements across populations. It was never designed to be an individual assessment, and it acquired the name "body mass index" only in 1972, when Ancel Keys tested several weight-for-height ratios and found this one correlated best with body fatness across large groups while being trivial to compute. That is the actual claim: cheap, and adequate across a population. It has been asked to do a great deal more work than that ever since, largely because it needs two numbers anyone can get in a doctor's waiting room.
Understanding that origin is the most useful thing on this page. A tool built to characterise a group, applied to one person, will be wrong about a lot of individuals in a way that is entirely consistent with it having been useful for the group. That is not a flaw in the calculation. It is what the calculation is.
Questions people ask
My BMI says overweight but I lift regularly. Is the number wrong?
The number is arithmetically correct and the label may still be misleading for you. BMI has no way to separate muscle from fat, so a well-muscled adult routinely reads in the overweight band with a body fat percentage in the low teens. This is the single best-documented failure mode of the index and it is why body composition estimates and waist measurement exist alongside it. If you want a second figure that responds to shape rather than mass alone, the tape-measure estimate on the body fat calculator uses waist, neck and height and will read very differently for you.
Which threshold applies to me, 25 or 23?
That is not something a calculator can settle. The lower cutoffs were introduced because studies in several Asian populations found the risk of type 2 diabetes and cardiovascular disease rising at lower BMI values than the standard bands imply, at least partly because body fat percentage tends to be higher at any given BMI. Whether they are the right frame for a given individual depends on ancestry, family history, waist measurement and blood work rather than on the index alone. Both are shown here so you can see the reading against each, and the choice between them belongs in a conversation with a clinician.
Does BMI work for children?
Not with these bands. Body composition changes substantially through growth, so a child's BMI is read against a percentile chart for their age and sex rather than against a fixed number. The same calculation is performed, but a BMI of 19 means something completely different for a 7-year-old, a 15-year-old and an adult. Use a paediatric growth chart, and take any interpretation from a paediatrician rather than from a threshold.
How much does BMI change from day to day?
Enough to make single readings noisy. Adult body weight swings by a couple of pounds across a day and across a week from food in the gut, fluid balance, glycogen, salt intake and, for people who menstruate, cycle phase. At 5 foot 10 a 3 lb swing is about 0.4 of a BMI point, which is more than enough to cross a category line and back. Weigh under the same conditions, first thing in the morning being the usual choice, and compare weekly averages rather than individual days.