Is BMI accurate? What it gets right, and the four cases it gets wrong.

BMI is the most cited screening number in medicine. It’s also the most misunderstood. Here’s the honest version: the formula is fine, the categories are arbitrary, and there are four well-characterised cases where the result is actively misleading for individual people.

What BMI is, and what it’s actually for

BMI is weight / height², in metric (kg / m²) or with the 703 conversion factor in imperial. Belgian statistician Adolphe Quetelet derived the index in the 1830s as a population-level descriptor. The American physiologist Ancel Keys popularised the modern name in a 1972 paper that compared height-weight indices for predicting cardiovascular disease in populations, not individuals.

This is the key distinction the public discourse usually misses. BMI was designed to characterise groups. Use it to compare the obesity prevalence of two countries, or to track the same population over decades, and it’s a workhorse. Apply it to a 33-year-old powerlifter and it fails on the first try.

Try yours in our BMI calculator — the result is the same number used in every clinical context. Read the rest of this guide to understand how to interpret it.

Where BMI works well

BMI is a reasonable screening signal for:

• Adults 20-65 with typical body composition. A sedentary office worker, a moderately active student, a retiree with moderate muscle mass — for these populations BMI correlates with body-fat percentage at roughly r=0.7. That’s not great, but it’s the kind of accuracy you can get with two numbers (weight + height) in fifteen seconds.
• Population-level public-health tracking. When you average across millions of people, individual mis-classifications wash out. The CDC’s obesity statistics, the WHO’s global health reports — these all use BMI for good reason.
• Tracking your own change over time.If your BMI moves from 27 to 24 over a year, the direction is meaningful even if the absolute numbers undersell the improvement (because BMI doesn’t distinguish muscle gain from fat loss).

The four cases where BMI misleads

1. The athlete problem (highly muscular adults)

Muscle is denser than fat — about 1.06 g/cm³ vs 0.92 g/cm³. At the same height, a person with high muscle mass weighs more than someone with high fat mass, and BMI scores them identically. The result: NFL linebackers, Olympic sprinters, and competitive weightlifters routinely classify as “overweight” or “obese” under WHO ranges while carrying single-digit body fat percentages.

What to use instead: Body-fat percentage via the US Navy tape method gets you within ±3% of DEXA at zero cost.

2. The sarcopenia problem (older adults)

Adults over 65 lose muscle at roughly 1-2% per year (a process called sarcopenia) while often maintaining or gaining body fat. Their BMI stays in the “healthy” range even as their actual body composition deteriorates. BMI in this group under-estimates health risk — the opposite of the athlete case.

What to use instead: Waist-to-height ratio (target: under 0.5) and grip strength testing. The latter requires a hand dynamometer.

3. The growth problem (children and adolescents)

Children’s body composition changes radically with development. A 10-year-old’s “normal” BMI is different from a 16-year-old’s, and from an adult’s. Clinical pediatric practice uses age- and sex-specific BMI percentile curvesderived from CDC growth-chart data — never the raw adult WHO bands. Plotting a 12-year-old’s BMI against the adult cutoffs will give you a meaningless number nine times out of ten.

Our BMI calculator surfaces this caveat on the page. For anyone under 20, use the BMI percentile tool your pediatrician uses — typically the CDC’s BMI-for-age calculator.

4. The ethnicity-blindness problem

The WHO category cutoffs (25 for overweight, 30 for obese) were calibrated against European-ancestry populations. South Asian and East Asian populations show meaningful cardio-metabolic risk at lower BMIs — type 2 diabetes prevalence rises sharply at BMI ≥ 23 in those groups, rather than 25.

Several national health bodies (Japan, India, China, Singapore) have adopted lower cutoffs for their populations. The WHO maintains the original universal cutoffs as the default and notes the Asian-specific guidance separately.

So what should you actually do?

For the typical healthy adult:

1. Compute your BMI in our calculator. Take note of which band you fall in.
2. Compute your body-fat percentage via the Navy method. This corrects for the athlete and sarcopenia cases.
3. Measure your waist (relaxed, at the navel). Divide by your height (same units). Target: under 0.5.
4. If all three line up — BMI in the healthy band, body-fat percentage in the healthy band, waist-to-height under 0.5 — you’re probably fine. If any two disagree, the body composition picture is worth a follow-up conversation with a clinician.

BMI’s mistakes aren’t a reason to discard it. They’re a reason to know when it’s talking out of its depth, and to keep one or two additional numbers within reach for those cases.

Worked example: three people at BMI 27

All three are 5'10" (178 cm) and 188 lb (85 kg), producing a BMI of 27.0 — clinically “overweight” under WHO bands.

• Person A:35-year-old recreational weightlifter. DEXA body fat: 14%. Waist 33", waist-to-height 0.47. Cardio-metabolic markers normal. BMI flags as overweight; every other signal says healthy body composition. The BMI reading is a false positive.
• Person B:35-year-old sedentary office worker. DEXA body fat: 28%. Waist 38", waist-to-height 0.54. Borderline triglycerides. BMI flags as overweight; the other signals concur. The BMI reading is correctly informative.
• Person C:75-year-old with mild sarcopenia. DEXA body fat: 34%. Waist 40", waist-to-height 0.56. Pre-diabetic. BMI of 27 is the “overweight” range — but the body composition is closer to clinical obesity than the number suggests. BMI is misleadingly reassuring.

Same BMI, three completely different clinical pictures. The number alone tells the physician very little about any specific patient; the secondary measurements (composition, waist circumference, lab work, activity level) are what actually inform care.

Common mistakes

• Treating a BMI jump as weight gain. BMI rises with both fat and muscle mass. Someone who started lifting weights and gained 10 lb of muscle will see their BMI rise; that’s not the same as a 10 lb fat gain. Body-fat percentage disambiguates.
• Using BMI for pregnant or post-partum women. Pregnancy invalidates the calculation (excess weight is fluid, baby, placenta — not composition). Post-partum BMI takes 6-12 months to stabilise as fluid shifts and breastfeeding resolves.
• Comparing BMI across very different heights. The weight/height² formula is a fair approximation for adults of typical height (150-190 cm) but breaks down at extremes. Very tall and very short adults are systematically mis-classified by the standard formula; the Trefethen index (weight/height^2.5) was proposed to correct this but is not in clinical use.
• Using BMI as a fitness measure. BMI is a body-composition proxy, not a fitness measure. Cardiorespiratory fitness (VO₂ max) and metabolic health are independent of BMI in many cases. A high-BMI individual with strong fitness has materially lower all-cause mortality than a normal-BMI sedentary individual.
• Targeting a specific BMI as a health goal. Optimising for a number is less useful than optimising for the underlying signal (composition, fitness, metabolic health). The BMI is downstream of those; it isn’t the lever.

For the broader composition story, see our BMI vs body-fat percentage guide. For the energy-balance question that often comes next, see how to calculate TDEE correctly.