BMI: The Formula, the Categories, and the Boundaries of the Metric

Body mass index looks like the most elementary calculation in health: divide weight by the square of height. The formula is one line long. Yet BMI remains one of the most misused numbers in public health, mostly because a single index is quietly doing several jobs — screening for population-level risk, tracking change over time, and being read as a statement about an individual's body composition. It is reliable for the first and acceptable for the second, and it does not do the third. Understanding what BMI measures, and what it deliberately ignores, is the difference between using it as a screening signal and mistaking it for a diagnosis.

The formula

With metric units, BMI is weight in kilograms divided by height in metres squared:

BMI = weight_kg / (height_m ^ 2)

With imperial units, the constant 703 makes the same calculation work directly in pounds and inches:

BMI = 703 × weight_lb / (height_in ^ 2)

Conversions are straightforward, and doing them before the division avoids most of the errors:

weight_kg  = weight_lb  × 0.45359237
height_m   = height_in  × 0.0254

The adult categories, as defined by the World Health Organization and adopted by most national health agencies, are:

BMI < 18.5        underweight
18.5 to 24.9      normal weight
25.0 to 29.9      overweight
30.0 and above   obesity

Some sources split the top band further into class I (30.0–34.9), class II (35.0–39.9), and class III (40.0+). The cut points are sometimes also shifted: many Asian countries use 23.0 as the overweight threshold and 27.5 as the obesity threshold, because risk at a given BMI rises at lower values in those populations.

For children and adolescents, BMI is interpreted against age- and sex-specific reference percentiles instead of fixed cut points: below the 5th percentile is underweight, the 85th to under the 95th percentile is overweight, and the 95th percentile or above is obesity. CDC and WHO publish separate reference curves, and their boundaries are not identical.

Worked example

A person is 1.75 m tall and weighs 78 kg.

BMI = 78 / (1.75 ^ 2) = 78 / 3.0625 = 25.47

That falls in the overweight band, just above the 25.0 cutoff.

The same person in imperial units: 1.75 m is 68.9 inches, and 78 kg is 172 lb.

BMI = 703 × 172 / (68.9 ^ 2)
    = 120,916 / 4,747.2
    = 25.47

The results match, as they should.

Now consider two people with identical BMI and different bodies. Person A is 78 kg and 1.75 m. Person B is 95 kg and 1.95 m:

B: BMI = 95 / (1.95 ^ 2) = 95 / 3.8025 = 24.98

B is in the normal band, is 17 kg heavier, and is 20 cm taller. A percentage of body fat at the same BMI differs substantially between the two, and the gap is roughly what separates a recreational runner from a sedentary desk worker at identical measurements.

The reverse case shows the metric's blind spot most clearly. A muscular athlete at 1.80 m and 88 kg has:

BMI = 88 / 3.24 = 27.16

Overweight by category, while carrying body fat well below average. BMI simply cannot see the difference between muscle and fat because it uses weight, and weight does not distinguish them.

Edge cases

• BMI is not a body composition measure. It cannot distinguish fat mass from lean mass, water, bone, or organ tissue. Two people at the same height and weight can have body fat percentages 15 points apart.
• No age or sex adjustment for adults. The adult cut points apply unchanged to everyone 18 and over, even though body composition and risk differ substantially between a 20-year-old and a 75-year-old at the same BMI.
• Pediatric use requires percentiles. A 10-year-old at BMI 19 and a 15-year-old at BMI 19 are in different categories; using adult thresholds for children produces frequent misclassification.
• Athletes and high muscle mass. Rugby players, weightlifters, and many endurance athletes routinely fall into the overweight band while being among the healthiest people measured.
• Older adults. The normal range shifts upward with age, since some loss of lean mass is expected, and the same BMI carries different risk at 70 than at 30.
• Ethnicity-specific risk. At the same BMI, risk of type 2 diabetes and cardiovascular disease is higher in South Asian populations, and BMI is generally less reliable as a risk predictor in these groups.
• Very short or very tall stature. The index is weight relative to height squared, and the relationship is not equally accurate at the extremes of the height distribution.
• Pregnancy. BMI categories do not apply; weight gain is assessed against trimester-specific gestational gain ranges.
• Edema, ascites, or fluid retention. Excess fluid inflates weight and therefore BMI without any change in body composition, which matters in heart failure, kidney disease, and advanced liver disease.
• Edible mass extremes. BMI is deliberately not used for underweight populations with nutritional scarcity, where it can misread malnutrition risk.
• Edible-cachexia and severe illness. Catabolic states change weight quickly, and BMI categories do not capture the rate or cause of the change.
• BMI is a screening tool, not a diagnostic test. Diagnosing obesity clinically requires additional measures: waist circumference, blood pressure, blood tests, and physical examination.

Common mistakes

• Treating a BMI band as a diagnosis. "You are obese" is not a clinical conclusion; the categories are population screening thresholds.
• Interpreting a category as a fixed personal risk. A BMI of 25.0 and one of 29.9 sit in the same band but are not the same risk, and risk also depends on everything BMI ignores.
• Using adult thresholds for children. The result is meaningless without age- and sex-specific percentiles.
• Confusing the overweight band with a lack of health. The band was drawn from population risk data, and individuals within a band vary enormously.
• Confusing BMI with body fat percentage. They are different quantities, and body fat percentage is a separate measurement with its own methods and error ranges.
• Mixing up metric and imperial inputs. Using pounds with a metres-based formula gives a BMI off by a factor of about 2.2, which lands many people in the wrong band.
• Using inches directly with the metric formula. Height must be converted to metres before squaring; substituting the raw inch value inflates BMI by about 2.7 times.
• Rounding height before converting. A height rounded from 173 cm to 170 cm changes BMI by roughly 1.5 percent, enough to cross a category boundary in some cases.
• Reading a category as a target. The bands were not designed as goals, and the "normal weight" band is not an ideal range for every individual.
• Ignoring the direction of change. A move from BMI 31 to 27 inside six months is meaningful regardless of which band each number falls in.
• Using BMI to judge a specific person. Two people with the same number can have different body composition and different health, and only the person and their clinician can say which applies.

Try it

The BMI calculator at https://www.miryori.org/tools/bmi-calculator handles both metric and imperial inputs, applies the adult category thresholds, and returns the value along with its category.

YMYL note: this is a mathematical estimate for general screening and is not medical advice, and a BMI result should not be used to diagnose, treat, or change any health decision without a qualified clinician.