DEXA is a bone scanner. It was built in 1987 to catch osteoporosis before a hip fracture did, and for most of its life that was the whole job. What it does now, almost as a side effect, is settle a much more everyday question: not whether your bones are thinning, but what the rest of you is actually made of.
That question has gotten louder as more people learn to distrust the bathroom scale and even BMI, which weighs a body but cannot see inside it. The honest next step is body composition: how much of you is fat, how much is muscle, and where it sits. The tools that answer this do not measure the same way, do not agree with each other at the individual level, and are not even fully consistent with themselves. Here is what each one is actually doing.
Two different physics, not two versions of the same test
A bathroom scale that prints a body-fat percentage, a gym's handheld grip device, and a machine like InBody are all running bioelectrical impedance analysis, or BIA. A weak electrical current passes through the body, and the device times how long it takes to get from one electrode to another. Current moves quickly through water and muscle and more slowly through fat, so the delay becomes an estimate of what you're made of. It is fast, cheap, and requires nothing more than standing on a plate or gripping a handle.
DEXA, or dual-energy X-ray absorptiometry, works on a completely different principle. Two X-ray beams at different energy levels pass through the body while you lie still for several minutes. Bone, fat, and lean tissue absorb each energy level differently, and a computer uses that difference to sort every pixel of the scan into one of the three. It is not estimating from a proxy signal the way BIA is; it is reading each tissue type more or less directly, which is part of why researchers treat it as the reference standard.
A bone scanner co-opted for fat
The history matters here because it explains why DEXA earned that trust in the first place, and it had nothing to do with body fat. Hologic introduced the first commercial DEXA system, the QDR-1000, in 1987, built by engineers David Ellenbogen and Jay Stein to diagnose osteoporosis faster and more precisely than the older isotope-based scanners it replaced. Body composition was a byproduct of the same X-ray physics, not the original design goal, and it took years of separate validation work before researchers trusted DEXA's fat and lean-mass readings enough to use them. By some accounts, DEXA has since replaced underwater weighing as the working reference method in body-composition research.
That pedigree is why the CDC chose DEXA to measure body composition in the National Health and Nutrition Examination Survey, the ongoing national health survey that has scanned tens of thousands of Americans since 1999 to build the reference curves clinicians and researchers use today.
Even the reference standard needed a correction
The part rarely mentioned in a DEXA sales pitch is that the "gold standard" has needed fixing before. A 2005 study in the American Journal of Clinical Nutrition, led by Dale Schoeller, found that one widely used Hologic scanner model systematically underestimated fat mass and overestimated lean mass compared with more rigorous multi-compartment reference methods. The fix wasn't cosmetic: NHANES now adds 5 percent of a scan's lean mass reading back onto its fat mass reading before publishing the number, a direct patch for an error the reference method itself was found to be making.
DEXA also isn't one single, interchangeable measurement. Scans from the two major manufacturers, Hologic and GE Lunar, do not report identical numbers for the same body, which is why researchers had to build cross-calibration equations just to make one machine's data comparable to the other's. In practice, the specific number a DEXA scan hands you is tied to the make, model, and software version of the machine you happened to lie down in, not to some fixed, universal truth about your body. Comparing scans from two different facilities can quietly compare two different rulers.
Where BIA actually stands
BIA's reputation problem is older and better documented. The National Institutes of Health convened a full technology assessment conference in 1994 specifically because of, in its own framing, the confusion surrounding how single- and multiple-frequency impedance readings should be interpreted. That confusion never fully resolved. A 2018 study comparing DEXA and BIA across 3,655 measurements found that agreement between the two methods was poor at the individual level regardless of a person's BMI, even though the two track each other reasonably well across large groups. A 2024 analysis of more than 34,000 UK Biobank participants scanned with both methods found the same pattern at larger scale: strong population-level correlation, but BIA underestimated fat mass by an average of 1.84 kilograms and overestimated fat-free mass by 2.56 kilograms, with the gap shifting by BMI and waist size. BIA is not wrong so much as it is a population-trained estimate applied to one individual body, and individual bodies are exactly where population assumptions break down.
Reading your own number
None of this means either tool is useless; it means each is suited to a different job. A DEXA scan is worth the cost and the drive when you want the most precise available answer at a single point in time, especially if you also care about the two things only DEXA reliably provides: regional detail, like how much fat sits viscerally around your organs, and bone mineral density itself. Book any follow-up scan at the same facility on the same machine, since switching brands can move the number for reasons that have nothing to do with your body.
A BIA device, whether a $30 smart scale or a gym's InBody station, is a reasonable tool for watching a trend over months, provided you hold the conditions steady: the same device, the same time of day, similar hydration and time since your last meal or workout. What it cannot do reliably is hand you an exact, standalone body-fat percentage you can trust down to the decimal, or one you can fairly compare against a friend's different scale or your own DEXA result from last year. Treat the shape of the trend as the signal, and treat any single BIA number as a rough estimate wearing the costume of precision.
Common questions
Is a home smart scale as accurate as a gym's InBody machine?
Both use the same underlying BIA principle, but InBody-style devices typically use more electrode contact points and multiple current frequencies, which tends to improve consistency. Neither should be read as an exact number, and switching between the two mid-tracking will make your trend line look like it moved when your body may not have.
How much does a DEXA scan cost, and does insurance cover it?
A body-composition DEXA scan booked purely for fitness or wellness purposes commonly runs $40 to
