Body Fat Percentage Estimation Formulas: BMI Method, Navy Circumference
Core Conclusion
No single estimation formula matches DEXA. Common field methods: BMI-derived (Deurenberg 1991) and US Navy circumference (Hodgdon-Beckett 1984). Both carry ±3–5% typical error versus reference methods. Criterion measurements include DEXA, hydrostatic weighing, and air-displacement plethysmography (Bod Pod).
Body fat percentage is among the most frequently requested body composition metrics in public health datasets, fitness assessments, and research protocols, yet it is also one of the most difficult to measure accurately outside of specialized laboratory settings. Unlike height and weight, which can be measured directly with basic equipment, body fat percentage requires either costly imaging and densitometry apparatus or statistical estimation from proxy anthropometric inputs. This article documents the technical definitions of the three principal laboratory reference methods, presents the full algebraic form of the two most widely cited field estimation formulas (the Deurenberg BMI-derived equation and the US Navy circumference method), reproduces the Gallagher et al. 2000 multi-ethnic age-category reference table, and summarizes published error ranges for each estimation technique against laboratory criterion measurements.
Readers seeking direct computation can use the [Body Fat Calculator + ../tools/body-fat-calculator.html] which implements both methods side-by-side with the same Siri/Brozek conversion chains used in the original papers. BMI values used as inputs to the Deurenberg formula can be computed separately in the [Adult BMI Calculator + ../tools/bmi-calculator.html].
Gold Standard Reference Method Definitions
All body fat percentage estimation formulas are validated against a criterion or reference method — a laboratory procedure considered sufficiently accurate to serve as the ground truth in comparative studies. The three principal reference methods currently in use are defined below in technical terms only.
Dual-Energy X-Ray Absorptiometry (DEXA / DXA)
DEXA uses two X-ray beams of different energy levels directed through the body. Bone mineral, lean soft tissue, and adipose tissue each attenuate the two X-ray energies by characteristic and distinguishable ratios. The DEXA scanner's software decomposes the total detected attenuation into three separate areal mass compartments: bone mineral content, lean soft tissue mass, and fat mass. Body fat percentage is computed as fat mass divided by total measured mass, multiplied by 100. DEXA measurements are typically performed in radiology clinics, sports science laboratories, or body composition research centers, and the procedure takes approximately 5 to 15 minutes of scan time per subject.
Hydrostatic Weighing (Underwater Weighing, Hydrodensitometry)
Hydrostatic weighing applies Archimedes' principle of buoyancy to estimate the average density of the human body. The subject is weighed on dry land and then submerged completely in a water tank of known temperature, with residual lung volume measured or estimated separately. The difference between dry weight and submerged weight, divided by the density of water at the measurement temperature, yields the body volume. Body density is then dry mass divided by body volume. A body-density-to-fat-fraction conversion equation (most commonly the Siri 1956 2-compartment equation or the Brozek 1963 update) converts the computed density into body fat percentage.
Air-Displacement Plethysmography (Bod Pod)
Air-displacement plethysmography uses air pressure and volume relationships in a sealed rigid chamber rather than water to estimate body volume. The subject sits inside the enclosed chamber. Pressure sensors measure the differential between an empty-chamber baseline and the with-subject condition. The chamber's known total volume minus the volume of air remaining in the chamber with the subject present gives the subject's body volume. As with hydrostatic weighing, the resulting body density is converted to body fat percentage using the Siri or Brozek equation, with a standard correction for estimated thoracic gas volume. The commercial trade name Bod Pod is the dominant implementation of this technique.
Deurenberg BMI-Derived Body Fat Formula (1991)
The Deurenberg equation was first published in 1991 by Paul Deurenberg, Marleen van der Kooy, and colleagues in the British Journal of Nutrition. The formula estimates body fat percentage as a linear function of three variables: BMI, age, and a binary sex indicator. The original validation sample was n=167 adult subjects, with hydrostatic densitometry used as the reference method. The authors reported a standard error of the estimate (SEE) of 3.9 percentage points of body fat[Deurenberg 1991].
Deurenberg Full Equation
Body Fat % = (1.20 × BMI) + (0.23 × ageyears) − (10.8 × S) − 5.4
Where:
- BMI = Body Mass Index in kg/m² (adult classification ranges)
- ageyears = Chronological age in completed years
- S = Sex constant: 1 for adult males, 0 for adult females
The coefficients reflect the statistical structure of the underlying population data. The positive BMI coefficient (1.20) captures the intuitive relationship between weight-for-height and adiposity at the population level. The positive age coefficient (0.23 per year) captures the age-related shift in body composition toward higher adiposity at identical BMI values, a pattern consistently documented in NHANES and other cross-sectional population datasets. The negative sex coefficient (−10.8) captures the average sex difference in body fat percentage at matched BMI and age.
Deurenberg and colleagues subsequently published modified and extended versions of the formula, including age-specific variants for pediatric and geriatric populations. The 1991 four-term linear form above remains the version most frequently reproduced in textbook and calculator implementations.
US Navy Circumference Method: Hodgdon-Beckett 1984
The US Navy circumference body fat estimation protocol was developed by James A. Hodgdon and Patricia B. Beckett at the Naval Health Research Center in San Diego, California. The full technical report was published in 1984. The protocol was designed for military population screening, where low equipment cost, quick administration, and standardized measurement technique were primary design constraints. The method estimates body density from height and a small number of circumference measurements, then converts the estimated density to body fat percentage using the Siri 1956 2-compartment conversion[Hodgdon-Beckett 1984].
Male US Navy Body Fat Formula (3 Sites: Height, Neck, Waist)
Body Densitymale = 1.0324 − 0.19077 × log10(waistcm − neckcm) + 0.15456 × log10(heightcm)
Body Fat %male = ((4.95 / Body Density) − 4.50) × 100 (Siri 1956 conversion)
Female US Navy Body Fat Formula (4 Sites: Height, Neck, Waist, Hip)
Body Densityfemale = 1.29579 − 0.35004 × log10(waistcm + hipcm − neckcm) + 0.22100 × log10(heightcm)
Body Fat %female = ((4.95 / Body Density) − 4.50) × 100 (Siri 1956 conversion)
Some implementations of the Navy method use the Brozek 1963 density-to-fat conversion ((4.570 / Body Density) − 4.142) × 100 instead of the Siri equation. The resulting numerical difference between Siri and Brozek conversions is typically on the order of 0.5 to 1.0 percentage point of body fat for most adult density ranges.
Male versus Female Input Differences
The male and female Navy circumference protocols differ in two structural ways. First, the female formula includes a hip circumference measurement that does not appear in the male formula. This additional input captures the sex-specific pattern of adipose tissue distribution in the derivation sample, where hip circumference was found to contribute statistically significant explanatory power for females but not for males.
Second, the two formulas differ in how the circumference variables are combined inside the logarithm term. The male formula computes waist circumference minus neck circumference. The female formula computes waist circumference plus hip circumference minus neck circumference. The base-10 logarithm of these circumference sums or differences then enters the linear body density equation with the coefficients shown above.
Gallagher et al. 2000 Body Fat Percentage by Age Category Table
Dympna Gallagher, Steven B. Heymsfield, Moonseong Heo, and colleagues published a multi-ethnic reference body fat percentile table in the American Journal of Clinical Nutrition in the year 2000. The study sample comprised 1,681 adult subjects (822 male, 859 female) spanning four self-identified ethnic groups (White, Black, Hispanic, Asian) and a wide age range (18 through 94 years). Reference measurements were collected by DEXA. Mean body fat percentage values are reproduced below in 5-year age bins[Gallagher 2000].
| Age Group (Years) | Males — Mean BF % | Females — Mean BF % |
|---|---|---|
| 20 – 24 | 22.0% | 33.0% |
| 25 – 29 | 23.5% | 34.2% |
| 30 – 34 | 24.6% | 35.1% |
| 35 – 39 | 26.0% | 36.3% |
| 40 – 44 | 27.5% | 37.2% |
| 45 – 49 | 28.8% | 38.0% |
| 50 – 54 | 29.8% | 38.6% |
| 55 – 59 | 30.4% | 38.9% |
| 60 – 64 | 30.8% | 38.8% |
| 65 – 69 | 31.0% | 38.7% |
| 70 – 74 | 31.2% | 38.8% |
| 75 – 79 | 31.5% | 39.1% |
Example Body Fat Calculations for Two Adults
Example 1 — Male, 35 years, 178 cm, 82 kg, BMI 25.88, Neck 40 cm, Waist 92 cm
Deurenberg BMI-derived method: BF% = (1.20 × 25.88) + (0.23 × 35) − (10.8 × 1) − 5.4 = 31.06 + 8.05 − 10.8 − 5.4 = 22.91%
US Navy circumference method (Siri conversion):
log10(92 − 40) = log10(52) = 1.7160
log10(178) = 2.2504
Body Density = 1.0324 − (0.19077 × 1.7160) + (0.15456 × 2.2504) = 1.0324 − 0.3273 + 0.3478 = 1.0529
BF% = ((4.95 / 1.0529) − 4.50) × 100 = (4.7013 − 4.50) × 100 = 20.13%
Example 2 — Female, 42 years, 166 cm, 70 kg, BMI 25.40, Neck 34 cm, Waist 80 cm, Hip 104 cm
Deurenberg BMI-derived method: BF% = (1.20 × 25.40) + (0.23 × 42) − (10.8 × 0) − 5.4 = 30.48 + 9.66 − 0 − 5.4 = 34.74%
US Navy circumference method (Siri conversion):
log10(80 + 104 − 34) = log10(150) = 2.1761
log10(166) = 2.2201
Body Density = 1.29579 − (0.35004 × 2.1761) + (0.22100 × 2.2201) = 1.29579 − 0.7617 + 0.4906 = 1.0247
BF% = ((4.95 / 1.0247) − 4.50) × 100 = (4.8307 − 4.50) × 100 = 33.07%
Typical Error Range Table by Method
| Body Fat Estimation Method | Reference Used | Typical SEE / Error Range | Validation Sample Size |
|---|---|---|---|
| Deurenberg BMI-derived (1991) | Hydrostatic Weighing | SEE ≈ 3.9 % BF | n = 167 |
| US Navy Hodgdon-Beckett Male (1984) | Hydrostatic Weighing | SEE ≈ 3.0 – 3.5 % BF | n = 384 (Navy male) |
| US Navy Hodgdon-Beckett Female (1984) | Hydrostatic Weighing | SEE ≈ 3.5 – 4.2 % BF | n = 207 (Navy female) |
| Skinfold 3-Site (Jackson-Pollock) | Body Density / Siri | SEE ≈ 3.5 – 5.0 % BF | Varies by study |
| Bioelectrical Impedance (Consumer Scales) | DEXA / Hydrostatic | SEE ≈ 3.8 – 6.0 % BF | Device-dependent |
| DEXA (Reference Criterion) | Direct X-ray measurement | ~1.0 – 2.0 % BF | N/A (Reference) |
The general pattern documented across validation studies is that all field estimation methods carry error on the order of ±3 to ±5 percentage points of body fat when compared to laboratory reference measurements. Error magnitudes tend to increase for individuals at the extreme ends of the adiposity spectrum, and for individuals whose body composition characteristics differ substantially from the derivation samples.