Durnin Womersley Protocol for Body Fat Estimation

The Durnin and Womersley protocol is a four-site skinfold method for estimating body fat percentage, developed in 1974 from measurements on 481 men and women between the ages of 16 and 72. It remains one of the most widely used skinfold approaches in clinical practice, fitness testing, and field research, largely because it requires only a skinfold caliper and a trained hand. A 2022 systematic review confirmed that the original equations are still valid for estimating body fat in adult populations, though their accuracy shifts depending on who you are measuring and who is doing the measuring.

What the Protocol Actually Involves

The protocol calls for skinfold measurements at four anatomical sites: the biceps (front of the upper arm), the triceps (back of the upper arm), the subscapular (just below the shoulder blade), and the suprailiac (just above the hip bone on the side of the torso). At each site, the tester pinches a fold of skin and subcutaneous fat between the thumb and forefinger, pulls it away from the underlying muscle, and applies the caliper jaws to get a thickness reading in millimeters. Each site is typically measured two or three times and averaged to reduce random error.

The four readings are summed, and that total is plugged into a logarithmic equation that produces an estimate of body density. Body density is then converted to body fat percentage using a secondary formula, most commonly the Siri equation or the Brozek equation, both of which assume a relatively constant density of fat tissue and fat-free tissue across individuals. The Durnin and Womersley equations are stratified by sex and age group, so you use different coefficients depending on whether you are measuring a 25-year-old woman or a 55-year-old man.

The choice of these four particular sites was deliberate. The original researchers wanted locations that could be measured on virtually anyone regardless of clothing or modesty concerns, that showed meaningful variation across body compositions, and that could be reliably found by testers with reasonable training. Compared to protocols that include thigh or abdominal skinfolds, the Durnin and Womersley sites tend to be easier to standardize across testers and are less affected by discomfort during measurement.

How Accurate Is It Compared to Lab Methods

The benchmark question for any body composition estimate is how closely it matches a criterion method, typically dual-energy X-ray absorptiometry (DXA) or hydrostatic (underwater) weighing. A 2022 systematic review of field-based body composition methods concluded that the Durnin and Womersley equation, along with the Jackson and Pollock equations, remains valid for estimating total body fat percentage in adult populations.¹ That does not mean the estimates are perfect. It means the protocol tracks DXA reasonably well on a group level and produces errors that are broadly acceptable for clinical screening and fitness assessment.

In a study comparing several skinfold equations against DXA in both African American and Caucasian American women, the Durnin and Womersley equation overestimated body fat by about 2.5 percentage points in African American women and about 2.7 percentage points in Caucasian American women.² Those are not trivial errors for an individual, but among the equations tested, Durnin and Womersley was consistently one of the least biased. The protocol tends to overestimate rather than underestimate, which means it errs on the side of telling you that you carry slightly more fat than you actually do.

In older adults, the picture is somewhat more encouraging. A comparison of prediction equations in non-obese older Caucasian Americans found that Durnin and Womersley had reasonably good agreement with DXA in older women, while Jackson and Pollock performed better in older men.³ Several competing equations significantly overestimated or underestimated body fat in these age groups, so the Durnin and Womersley protocol held up relatively well against the field.

Where Racial and Ethnic Differences Matter

The original 1974 data came predominantly from white British subjects, and that sample composition has consequences for how well the equations generalize. A large revisiting of the protocol using DXA as the reference found that the equations predicted body fat within about one percentage point for Caucasian men and African American women, but significantly overestimated it in several other groups. Asian women and African American men showed the largest discrepancies, with overestimates of roughly 3.3 and 2.4 percentage points respectively.⁴ The age-specific versions of the equations overestimated body fat across all groups tested.

The same study developed updated regression equations specific to each racial and ethnic group, incorporating body weight, height, and waist circumference alongside the original four skinfold sites to improve accuracy.⁵ These updated equations have not replaced the originals in most practical settings, partly because the original protocol’s simplicity is its selling point and partly because many practitioners are unaware the updates exist. If you are using the original equations on a diverse population, you should expect some systematic bias, particularly in groups that were underrepresented in the 1974 sample.

Performance in Athletes and Lean Individuals

Skinfold protocols in general tend to struggle with very lean or very muscular bodies, because the relationship between subcutaneous fat and total body fat changes as you move toward the extremes. The Durnin and Womersley protocol, however, has shown surprisingly decent performance in at least some athletic populations. In a study of elite sport climbers, the Durnin equation was the most accurate of 17 tested equations for estimating body fat percentage in both men and women, with an inter-method difference of less than one percentage point compared to DXA in males.⁶ In female climbers, the limits of agreement were tighter with Durnin’s equation than with competing formulas. The researchers noted the sample was not randomly selected, so the results should be taken as encouraging rather than definitive, but it does suggest the protocol is not as poorly suited to lean athletes as some practitioners assume.

That said, there are contexts where the protocol clearly falls short. In formerly obese adults who had lost weight, Durnin and Womersley significantly overestimated body fat compared to hydrostatic weighing, with a mean difference of about 7 percentage points and a total error of 8.4 percentage points.⁷ The Jackson-Pollock equation performed far better in this population, with a mean difference under one percentage point. The likely explanation is that people who have lost large amounts of weight carry their remaining subcutaneous fat differently, and the relationship between skinfold thickness and total body density that Durnin and Womersley assumed no longer holds. If you are testing someone who has undergone major weight loss, the protocol is not the right tool.

Special Populations and Unusual Contexts

The protocol has been tested in populations far outside its original scope. A study of Thai individuals with chronic traumatic spinal cord injury found that the Durnin-Womersley formula based on the four-site skinfold was a more practical method for evaluating obesity than other anthropometric approaches available in that population.⁸ In settings where access to DXA or other laboratory equipment is limited, skinfold measurement may be the only realistic option, and Durnin-Womersley’s simplicity gives it an edge over protocols requiring more sites or more complex calculations.

In a comparison among Pakistani adults, where the Durnin and Womersley, Jackson and Pollock, and Sloan equations were all tested against bioelectrical impedance analysis, all three underestimated body fat percentage. The Durnin and Womersley equation came closest to the reference values and showed the strongest correlation.⁹ It is worth noting that bioelectrical impedance is itself an estimation method, not a true criterion like DXA. But in field settings where neither DXA nor underwater weighing is available, the study suggests that Durnin and Womersley is a reasonable alternative.

Caliper Choice Is Not Trivial

One factor that practitioners often overlook is the caliper itself. The physical and mechanical characteristics of different calipers produce systematically different skinfold readings, even in the same person measured at the same site. A 2023 review concluded that the considerable variety of original and generic caliper types available is a source of systematic measurement error, and that the interchangeable use of different calipers is effectively impossible.¹⁰ The Harpenden caliper is traditionally considered the reference standard and was the type used in many of the original validation studies. Less expensive plastic calipers, commonly used in gyms and student labs, often apply different jaw pressures or have different jaw face areas, which can shift readings by several millimeters.

The practical takeaway is straightforward: if you are tracking body composition changes over time using skinfolds, use the same caliper every time. Switching from a Harpenden to a Slim Guide to a generic plastic caliper between assessments introduces error that can easily dwarf the actual change in body fat you are trying to detect. The protocol’s equations were developed using specific caliper types, and while many calipers produce correlated readings, they are not interchangeable in absolute terms.

Tester Skill and Measurement Consistency

The human element is at least as important as the hardware. Skinfold measurement is a manual skill, and the consistency of the person wielding the caliper determines how much noise enters the data. A study using accredited anthropometrists from the International Society for the Advancement of Kinanthropometry (ISAK) found that the average intra-evaluator technical error of measurement for skinfolds was about 1.15%, while the inter-evaluator error was roughly 2.34%.¹¹ These were highly trained measurers. In less experienced hands, the variability can be substantially larger.

This has a direct implication for anyone using the Durnin and Womersley protocol for serial testing: the same person should perform all measurements on a given client. The within-tester consistency is roughly half the between-tester variability, so switching testers between assessments creates apparent changes in body fat that are really just measurement noise. In a research setting, this means training and calibrating testers against each other before data collection. In a gym or clinic setting, it means assigning one staff member to a client’s body composition assessments and sticking with that pairing.

Does Hydration Affect the Results

A common concern is whether being dehydrated or overhydrated before a skinfold test distorts the numbers. The evidence is mixed but generally reassuring. A study examining exercise-induced dehydration in athletes found that while individual skinfold thicknesses at some sites increased after exercise (likely due to redistribution of fluid and increased skin blood flow), the effect on overall compressibility was limited.¹² A separate study in wrestlers found that body fat estimates from skinfolds were not significantly different across dehydrated, rehydrated, and normally hydrated states, and that drinking about a liter of water was enough to restore normal hydration without affecting the body fat percentage estimate.¹³

This is an area where skinfolds actually have an advantage over bioelectrical impedance, which is much more sensitive to fluid shifts. If you are testing athletes who have just finished training, or individuals who may not be in a perfectly controlled hydration state, skinfold-derived estimates are more robust than impedance-based methods. That said, best practice still recommends testing under consistent conditions: same time of day, similar hydration status, and not immediately after intense exercise.

How It Stacks Up Against Jackson and Pollock

The most common comparison in the literature is between Durnin and Womersley and the Jackson-Pollock equations, which use three or seven skinfold sites depending on the version. Neither protocol is categorically superior. The 2022 systematic review that validated both noted that Jackson and Pollock (three-site version for men) and Jackson, Pollock, and Ward (three-site version for women) are valid alternatives.¹⁴ In formerly obese adults, Jackson-Pollock handily outperformed Durnin-Womersley.¹⁵ In older women, Durnin-Womersley had the edge.¹⁶ In lean athletes, Durnin-Womersley outperformed a large array of alternatives.¹⁷

The Jackson-Pollock three-site protocol includes the chest, abdomen, and thigh in men, which can be harder to standardize and may involve more modesty-related logistical concerns, especially in clinical settings. The Durnin and Womersley sites are somewhat more accessible and less invasive. In practice, many professionals pick the protocol they were trained in and stick with it, which is a defensible approach as long as they understand the limitations of whichever set of equations they are using.

When to Use Something Else Entirely

The Durnin and Womersley protocol is a reasonable choice for general adult populations in settings where laboratory equipment is not available. It works well for tracking changes over time in the same individual when the same tester and caliper are used. It is probably the simplest of the validated skinfold protocols, which makes it a natural default for field work, fitness testing, and clinical screening in resource-limited environments.

There are clear situations where you should reach for a different method. If your subject has a BMI above 35 or so, skinfold measurement becomes mechanically difficult: the caliper jaw may not open wide enough, the fold may be hard to separate from underlying tissue, and the equations were not developed for that range of subcutaneous thickness. Formerly obese individuals who have lost large amounts of weight are also poorly served by this protocol, as the altered distribution of subcutaneous versus visceral fat breaks the assumptions baked into the density conversion. For these populations, DXA or air displacement plethysmography will give you meaningfully better data if they are available.

For research requiring individual-level precision rather than group-level estimates, skinfolds of any kind are generally insufficient. The standard error of estimate for skinfold-derived body fat percentage, even with a well-validated equation and a skilled tester, is typically in the range of three to four percentage points. That is fine for screening, for tracking trends, and for classifying people into broad categories. It is not fine for detecting a two-percentage-point change in body fat with confidence in a single person over a short time frame. If that level of precision matters, you need a criterion method.

Getting the Most Out of the Protocol in Practice

If you are going to use Durnin and Womersley, a few practical details make the difference between a useful measurement and a noisy one. Measure on the right side of the body (the original study used the right side, and switching sides introduces variability). Take each skinfold at least twice, and if the two readings differ by more than about a millimeter, take a third and average. Apply the caliper about one centimeter below where your fingers are pinching, wait two to three seconds for the reading to stabilize before recording, and release the skinfold between repeated measurements to allow the tissue to recover its shape.

Mark the sites with a skin-safe marker if you are new to the protocol. The subscapular site in particular can be tricky to locate consistently: it sits just below the inferior angle of the scapula, at an angle running diagonally down and outward. The suprailiac site varies between protocols; Durnin and Womersley specified a site above the iliac crest in the midaxillary line, but some practitioners confuse it with the supraspinale site used in other systems, which sits slightly forward and lower. Using the wrong landmark shifts the sum of skinfolds and cascades into the final estimate.

Record the raw skinfold values, not just the final body fat percentage. Raw values let you or a colleague recalculate using different equations if needed, spot which sites are driving changes over time, and compare across protocols. The percentage is the output most people care about, but the raw data is where the information actually lives.