BMI vs body fat percentage: which one should you actually track?

BMI (Body Mass Index) and body fat percentageare two completely different measurements that get used interchangeably in fitness culture. BMI is weight relative to height. Body fat percentage is, well, the fraction of body mass that’s fat. A muscular athlete can have a BMI of 28 (“overweight”) and 12% body fat (lean). A sedentary office worker can have a BMI of 24 (“healthy”) and 28% body fat (high). The numbers are measuring different things; the right tool depends on what you’re trying to learn.

What each one measures

BMI

BMI = weight (kg) / height² (m²). Devised by Adolphe Quetelet in the 1830s as a population statistic. Categories per WHO:

• < 18.5 — underweight
• 18.5-24.9 — healthy
• 25.0-29.9 — overweight
• ≥ 30.0 — obese

BMI is a screening tool, not a diagnosis. It correlates with body fat at the population level (heavier people, on average, have more body fat) but is unreliable for any specific individual. It systematically misclassifies muscular people as overweight and skinny-fat people as healthy.

Body fat percentage

The fraction of body mass that’s adipose tissue. Healthy ranges per American Council on Exercise:

Measurement methods, ranked by reliability

1. DEXA scan — gold standard. ±1-2% accuracy. Requires a clinical or specialised facility; $50-150 per scan in the US.
2. Hydrostatic weighing— “dunk tank.” ±2-3% accuracy. Increasingly rare; mostly replaced by DEXA.
3. Bod Pod — air displacement. ±2-3%. Found at sports science labs and a few high-end gyms.
4. Skinfold caliper — measured by a trained tester. ±3-5% if done correctly; ±10% if done badly. Free at most gyms. Highly tester-dependent.
5. Bioelectrical impedance (smart scales, handheld devices) — ±4-8% on a typical home device. Hydration state, recent meals, and time of day all shift the reading.
6. Navy circumference method— neck, waist, and (for women) hip measurements plugged into a formula. ±3-4% if the measurements are precise. Used by US military for compliance because it’s cheap and standardised.
7. Online calculators / BMI-derived estimates — convert BMI + age + sex to body fat. ±5-8%. Better than nothing if you have only BMI to start with.

When to use which

BMI is the right tool when:

• You’re tracking a large population (epidemiology, clinical screening).
• You don’t have access to a body-fat measurement.
• You’re looking for a quick first-pass health signal.
• You’re tracking weight change over time and your body composition isn’t changing dramatically (no recent muscle gain/loss).

Body fat percentage is the right tool when:

• You’re an athlete or training seriously for strength/size.
• BMI says you’re overweight but you have visible musculature.
• BMI says you’re healthy but you suspect “skinny fat” (low muscle mass, high relative fat).
• You’re cutting or recomposing and want to know whether weight loss is fat or muscle.

The pragmatic combination

For most people, the useful workflow is:

1. Start with BMI. Free, instant, gives you a category. Use our BMI calculator.
2. If BMI is borderline (24-27) or you have an athletic build, measure body fat. Skinfold or Navy method at minimum; DEXA if you’re serious. Our body fat calculator implements the Navy method.
3. Track waist circumferencealongside either number. Waist correlates with visceral fat (the metabolically dangerous kind) better than BMI does. WHO thresholds: men > 102 cm and women > 88 cm indicate elevated risk regardless of BMI.
4. If you’re training for body composition, repeat measurements every 6-12 weeks. Body fat changes slowly; daily measurements add noise without signal.

Common mistakes

• Smart-scale body fat as ground truth. Bioimpedance on a foot-to-foot scale reads body fat in the lower body only and extrapolates. ±5-10% error is normal. Useful for trend if measured under identical conditions (same time of day, same hydration), useless as an absolute number.
• Treating BMI categories as diagnoses. BMI 25 is not a disease. Many people with BMI 26-28 have excellent metabolic health. Conversely, BMI 22 with poor diet and no exercise carries real health risks.
• Comparing measurements across methods. A DEXA reading of 14% and a smart-scale reading of 20% on the same person on the same day is normal. Don’t compare numbers from different methods.

Worked example: when the two numbers disagree

A 38-year-old male amateur powerlifter: 178 cm, 92 kg.

• BMI: 92 / 1.78² = 92 / 3.168 = 29.0 — overweight, one step from obese. Per BMI alone, his GP might counsel weight loss.
• Navy method: neck 41 cm, waist 84 cm. Plug into the Navy equation: 495 / (1.0324 − 0.19077·log₁₀(84 − 41) + 0.15456·log₁₀(178)) − 450 ≈ 14% body fat — squarely in the “Athlete” band.
• Waist circumference: 84 cm, well below the 102 cm risk threshold from the NIH/Jensen 2014 criteria.
• Reconciliation:the BMI signal is a false positive caused by muscle mass. Body-fat percentage and waist confirm low metabolic risk. The right counsel is “continue training; weight loss not indicated.” The opposite case — BMI 22, body fat 30%, “skinny fat” — is the false negative; same person needs strength training even though BMI looks healthy.

Edge cases where the standard rules fail

• People of South or East Asian descent.WHO-Asia recommends lower BMI thresholds: overweight ≥23, obese ≥27.5 (vs the Western 25 / 30). The same BMI carries more diabetes risk in these populations.
• Older adults (65+). Slightly higher BMI (25-27) correlates with bettersurvival in longitudinal studies — the “obesity paradox.” Aggressive weight loss in this age group risks sarcopenia and increased fall mortality.
• Pregnancy.BMI categories don’t apply; use pre-pregnancy BMI for gestational weight-gain targets per IOM 2009 guidelines.
• Amputees and people with limb-length asymmetry. BMI denominator assumes both legs; calculated values are systematically biased. DEXA or bioelectrical impedance with a corrected formula gives better numbers.
• Bodybuilders in show-prep.Body fat <5% in men or <12% in women is not a health goal — it requires hormonal and metabolic adaptations that aren’t sustainable beyond a few weeks.

Common mistakes

• Comparing two body-fat readings from different methods. DEXA reads visceral fat directly; bioimpedance estimates total water and infers fat; skinfolds read subcutaneous fat only. Day-of differences of 5-8 percentage points between methods are routine and not informative about real change.
• Daily weighing as a body-composition signal.Day-to-day weight bounces ±1-2 kg from glycogen, sodium, and bowel content. Use 7-day rolling averages or fortnightly snapshots.
• Calculating BMI from imperial weights without unit conversion. BMI in metric is kg / m²; in imperial it’s 703 × lb / in². The 703 constant is easy to forget. Use a calculator.
• Treating “overweight” as a unified category. BMI 25-27 with high muscle, low waist, and normal blood pressure has a very different risk profile from BMI 27-30 with high waist and family history of diabetes. Two screening tools agreeing is more meaningful than one alone.

Sources: World Health Organization, “Body Mass Index - BMI” (Regional Office for Europe, 2024); American College of Sports Medicine, Guidelines for Exercise Testing and Prescription (11th ed., 2022); Jensen MD et al.,Circulation 2014 (waist circumference thresholds); NHLBI Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults.