Is a DEXA Scan Accurate? Factors Affecting Your Results

DEXA scans are among the most widely used tools for measuring bone density and body composition, but they are not perfectly accurate. Compared to the gold-standard four-compartment model for body fat, DEXA readings can differ by several percentage points, and for bone density, overlying calcifications and positioning errors can shift results enough to change a clinical diagnosis. The technology is reliable enough for most clinical and fitness purposes, yet a surprising number of everyday variables influence what the machine reports, from how hydrated you are to which brand of scanner the clinic uses.

How the Technology Works, in Brief

A DEXA scanner passes two low-dose X-ray beams at different energy levels through your body. Dense materials like bone absorb more energy than soft tissue, and fat absorbs differently than lean mass. The machine calculates a ratio of how much each beam was weakened as it passed through you. That ratio stays consistent for bone and fat across all people, but the ratio for soft tissue shifts depending on your individual mix of lean and fat mass, which is how the scanner estimates your body composition region by region.¹ This two-beam approach is the reason DEXA is called “dual-energy,” and it’s also the source of many of the scan’s limitations: anything that changes the thickness or density of the tissue the beams pass through can throw off the math.

How DEXA Compares to More Precise Methods

No body-composition method is perfect, but researchers treat the four-compartment model as the closest thing to a true answer. It combines underwater weighing, a measure of total body water, DEXA, and other data to separate your weight into fat, water, mineral, and protein. When DEXA body-fat readings were compared against this standard in one study, DEXA underestimated body fat by about two percentage points on average, with individual differences ranging from roughly negative three to positive seven percent.² Leaner people tended to see the biggest underestimates, likely because of a phenomenon called beam hardening, where differences in tissue thickness alter how the X-rays behave. In older adults, the disagreement was even wider, with a total error of about five percentage points against the four-compartment model.³ In children, DEXA also showed a measurable bias in estimating fat mass compared to the four-compartment approach.⁴

This doesn’t mean DEXA is useless. For tracking changes over time and for clinical decision-making about bone health, it remains the standard tool. But if you’re chasing a precise body-fat number for personal fitness goals, it helps to know the reading might be off by a few points in either direction, depending on your body type and age.

Hydration Can Shift Your Lean Mass Reading Substantially

Your hydration state is one of the biggest and most underappreciated sources of error. DEXA cannot tell the difference between water in your muscles and actual muscle tissue; both show up as lean mass. In a study that deliberately manipulated hydration, participants who exercised enough to lose about two kilograms of body weight through sweating saw their measured lean mass drop by roughly 1.7 kilograms. When those same participants then over-hydrated, gaining about 2.5 kilograms of water weight, their lean mass reading jumped by about 2.4 kilograms. Fat mass and bone mineral content didn’t change at any point, confirming the issue was entirely about water being misclassified as lean tissue.⁵

The practical takeaway is straightforward: if you get scanned after a hard workout on a hot day, you’ll look like you lost muscle. If you load up on water and carbs (which pull water into muscles) the night before, you’ll look like you gained it. Neither reading reflects a real change in your body composition.

Recent Exercise Muddies the Numbers Too

Even setting hydration aside, an exercise session before your scan introduces noise. Working out causes fluid shifts, changes blood volume distribution, and may prompt eating and drinking that further alter your tissue state. Research on active people found that exercise and its related fluid and food intake changed average body-composition estimates by amounts ranging from trivial to small but meaningful, and increased the typical measurement error by about ten percent.⁶ The simplest fix is to show up fasted and rested.

Positioning Matters More Than You’d Think

When you lie on the scanner table, small differences in how your body is arranged can change the results. Something as minor as where your hands are placed affects both bone density and body composition readings. Placing the hands in a mid-prone position (palms partially facing down) rather than flat increased total-body bone density readings and shifted lean and fat mass estimates.⁷ The effect happens because the position changes how much bone area the scanner sees, which cascades through the math for everything else.

The larger positioning problem is inconsistency between technologists. A study that audited scan positioning at the lumbar spine and hip found improper positioning rates of about 49% and 57%, respectively, before technologists were retrained. After retraining, those rates dropped to single digits, and the precision of successive measurements improved.⁸ In other words, roughly half of all scans at one center were positioned incorrectly before anyone noticed. If your bone density results seem surprisingly good or surprisingly bad, the technologist’s technique is worth considering.

Operator Skill and Region-of-Interest Placement

After the scan is taken, software automatically draws boundaries around body regions — your arms, legs, trunk, and so on. These boundaries aren’t always placed correctly by the algorithm, and how the operator adjusts them matters. In one reproducibility study, automatic placement of arm-region boundaries produced correlation values as low as 0.74 for repeated lean-mass measurements, meaning the software’s estimates bounced around quite a bit. When a trained operator manually adjusted those regions, the correlation jumped to 0.93.⁹ Total-body results tend to be more consistent than regional ones, so if a clinic tells you that your right arm lost lean mass but your left didn’t, treat that with some skepticism unless the scan was carefully reviewed.

Different Machines Give Different Numbers

Not all DEXA scanners agree with each other, and the two dominant manufacturers produce results that can’t simply be swapped. When the same group of adults was scanned on both a GE Lunar Prodigy and a Hologic Horizon, the GE system reported significantly higher appendicular lean mass — about 1.8 kilograms more in men and 1.0 kilogram more in women. The gap was even wider in people at the extremes of body weight or body fat.¹⁰

A separate comparison of a GE Lunar iDXA and a Hologic Horizon A found that the GE system reported about 3% less total lean mass and about 23% more visceral fat than the Hologic, while total fat mass was similar between them.¹¹ A large multinational standardization effort found average differences of about 1.4 percentage points for body fat, 177 grams for bone mineral content, and 0.013 g/cm² for bone density between GE and Hologic systems.¹² Cross-calibration equations exist to translate between brands, but the bottom line is that you should always use the same machine for follow-up scans if you’re tracking changes over time. Comparing a reading from a GE scanner six months ago to a Hologic scanner today is comparing apples to oranges.

Visceral Fat Estimates Are Especially Rough

Many newer DEXA machines advertise a visceral adipose tissue (VAT) reading — the deep abdominal fat that wraps around your organs and is linked to metabolic disease. This feature sounds impressive, but when DEXA visceral fat estimates were compared against MRI (the gold standard for imaging internal fat deposits), DEXA overestimated visceral fat volume at baseline by about 13%. More problematically, when tracking changes over time, DEXA underestimated the actual amount of visceral fat lost by roughly 33% at three months and 47% at twelve months.¹³ If you’re using DEXA to judge whether a diet or exercise program is reducing your visceral fat, the scan may dramatically understate your progress.

Body Size at the Extremes

DEXA scanners have a fixed scanning field, and they work best on people whose bodies fall within a moderate range. In severely obese individuals, results start to diverge from other methods. One study found significant differences between DEXA and underwater weighing for body-fat percentage in people with a mean BMI around 36, while the same methods agreed well in those people after they lost about 14% of their body weight.¹⁴ The issue is partly about beam hardening through thicker tissue and partly about the physical challenge of fitting someone on the scanning bed so that all tissue is captured. Some facilities use half-body scan protocols and double the result for the other side, which introduces its own set of assumptions.

Greater body size also erodes the precision of bone density measurements specifically. The “least significant change” is a threshold that tells clinicians how much a bone density reading needs to shift between scans before they can be confident the change is real and not just measurement noise. Research has shown that this threshold rises as obesity increases, meaning it takes a bigger swing in the number before doctors can be sure anything actually changed.¹⁵

Added soft tissue thickness reduces measured bone density directly. Layering even a modest amount of extra soft tissue over cadaveric bone specimens lowered density readings by 1% to nearly 4%, with greater errors at higher actual bone densities.¹⁶ For someone who gains or loses a significant amount of weight between scans, this means the apparent bone density change is partly a soft-tissue artifact.

Artifacts That Can Fake Higher Bone Density

DEXA measures density by looking straight through you from front to back. Anything dense that sits in the path of the X-ray beam gets counted, whether it’s bone or not. Calcified aortic plaque overlying the lumbar spine is one of the most common culprits. In one documented case, a calcified aorta contributed up to 33% of the measured bone density at the mid-lumbar region, which could easily push someone from an osteoporosis diagnosis into a “normal” range that doesn’t reflect their actual fracture risk.¹⁷

Degenerative spinal changes — bone spurs, disc calcification, collapsed vertebrae — create similar problems. Postmenopausal women with visible spinal degenerative calcification had higher spine bone density by the standard dual-photon technique, but when measured by a method that can separate out overlying structures, the difference vanished, confirming the elevation was an artifact.¹⁸ This is why hip measurements are often considered more reliable for older adults: the hip joint is less affected by vascular and degenerative calcification than the lumbar spine.

How DEXA Stacks Up Against BIA and Bod Pod

If DEXA has these limitations, you might wonder whether cheaper methods like bioelectrical impedance analysis (BIA, the technology in smart scales and handheld devices) or the Bod Pod (which measures body volume using air displacement) are just as good. In obese adults, two segmental BIA devices and the Bod Pod both produced body-fat estimates close to DEXA on average, with standard errors under about 2.5 percentage points for BIA and 2.4 for the Bod Pod. However, all methods showed proportional bias, meaning they performed less consistently at the extremes of body fatness.¹⁹ A separate study found that the Bod Pod overestimated body fat in thinner people and underestimated it in heavier people compared to DEXA, with some individual differences reaching 13 percentage points in underweight participants.²⁰

DEXA’s advantage over these alternatives isn’t necessarily that it’s more accurate in an absolute sense, but that it provides regional breakdowns (arm vs. leg vs. trunk) and bone data that the others can’t. For tracking overall trends, a good BIA device used consistently under the same conditions will show you the direction of change reasonably well. It just won’t give you the granular picture that DEXA offers.

Getting the Most Out of a Follow-Up Scan

Since so many variables can shift DEXA results, researchers who track body composition in athletes and clinical patients have developed standardization protocols. The core recommendations are consistent across multiple reviews: use the same machine, the same clinic, and ideally the same technologist every time.²¹ Phantom calibration scans (scanning a known block of material to verify the machine hasn’t drifted) should be performed routinely.²²

On the patient side, preparation matters. One well-designed study defined a standardized pre-scan protocol that required participants to fast for at least eight hours, avoid caffeine, nicotine, and exercise during that window, and arrive in the morning.²³ You don’t have to follow an extreme regimen, but a few rules help enormously:

• Same time of day: Morning, fasted, is the most reproducible window.
• Same hydration routine: Don’t water-load the night before one scan and dehydrate before the next.
• No hard exercise: Avoid strenuous training for at least 12 hours prior.
• Same clothing: Wear light athletic clothes without zippers, snaps, or metal.

Following these steps won’t eliminate measurement error, but it shrinks the biological noise enough that a real change is more likely to stand out from the background variability.

Special Considerations in Children

DEXA interpretation gets particularly tricky in pediatric patients. The scan measures bone mineral content and areal bone density — essentially, how much mineral is packed into a flat projection of the bone. In a growing child, a shorter-than-average kid will appear to have lower bone density simply because their bones are smaller, not because their bones are weaker. Conversely, a tall child may look falsely reassuring. Without height correction, clinicians risk diagnosing osteoporosis in short healthy children and missing it in tall ones.²⁴ There’s no universal consensus on the best correction method, though several approaches using height-for-age adjustments have been proposed.²⁵ If your child has had a DEXA scan, it’s worth asking whether the reading was adjusted for their size.

Reference Populations and What “Normal” Means

When a DEXA report gives you a Z-score or a percentile rank, it’s comparing you to a reference database. That database was built from a particular population, and if your body doesn’t match that population’s demographics well, the comparison may be misleading. Data from the Bone Mineral Density in Childhood Study showed meaningful differences across racial and ethnic groups in body composition indices, with Asian participants scoring on average about half a standard deviation below the mean for appendicular lean mass while Black participants scored about three-quarters of a standard deviation above it.²⁶ These are not differences in actual health status but in what the reference curve considers “average.” If your report flags you as low in lean mass or high in fat mass, it’s worth considering whether the reference population matches your demographics before you worry.

Radiation Exposure

One reason DEXA remains popular despite its accuracy limitations is that the radiation dose is extremely low. For context, the worldwide average background radiation everyone receives just from living on Earth is about 2.4 millisieverts per year.²⁷ A single DEXA scan delivers a tiny fraction of that — typically in the range of a few microsieverts, comparable to the background radiation you absorb in a few hours of normal life. This makes it safe enough for repeated scans, which is important because tracking changes over time is where DEXA provides the most clinical value. If your concern is whether getting two scans a year poses any meaningful radiation risk, the answer is that it does not.