What each method actually measures
The DEXA vs InBody question is really a question about two different physical principles, and the difference explains everything else that follows.
A DEXA scan passes two low-energy X-ray beams through the body. Bone, lean tissue, and fat absorb those beams differently, so the scanner can separate the three and map each across the body, region by region. It measures composition rather than inferring it.
An InBody sends a small, safe electrical current through the body and records the resistance it meets. Lean tissue holds water and conducts well; fat resists. From that resistance the device estimates total body water, then converts that estimate into fat and lean mass using the manufacturer's own equations. The measurement is real; the body fat figure is a prediction built on top of it.
| DEXA | InBody | |
|---|---|---|
| Principle | Dual-energy X-ray absorption | Bioelectrical impedance |
| Fat estimate | Measured directly | Predicted from body water |
| Bone density | Yes | No |
| Regional breakdown | Fat and lean, limb by limb | Lean mass by segment |
| Time | Roughly 7–20 minutes | Under a minute |
| Radiation | Very low dose | None |
| Repeatable often | Practical, not casual | Yes |
DEXA and InBody at a glance.
Different Health measures body composition in-lab with an InBody 770 for exactly the reason that table implies: it is fast enough and consistent enough to repeat at every visit, which is what makes a trend visible. For anyone who wants the more precise single read, a DEXA scan can be arranged.
How accurate is each one
Most comparisons stop at calling DEXA the gold standard. The published validation data is more specific and more useful than that.
| Finding | Sample | Result |
|---|---|---|
| InBody repeatability | 31 men, 36 women, InBody 230/720/770 [1] | ICC ≥0.98 for body fat %; standard error of measurement 0.77–0.99 percentage points |
| InBody vs DXA agreement | 480 men, 315 women, InBody 770 vs GE Lunar iDXA [2] | Mean absolute error 3.9 percentage points for body fat %; r² = 0.96 |
| DEXA same-day precision | Repeat scans with repositioning [3] | Precision error under ~1% for total body masses |
| DEXA across days | 21 resistance-trained athletes [4] | Consecutive-day precision error for fat mass nearly double same-day: 1,261 g vs 660 g |
| DEXA over months | Weight-stable adults, GE Lunar iDXA [5] | Coefficient of variation 0.8–5.9% over three months, versus under 2% same-day |
What validation studies report for each method. Precision means agreement between repeat measurements; accuracy means agreement with a reference method.
Two things in that table are worth sitting with. InBody's repeatability is genuinely excellent, better than a percentage point. And DEXA's precision is not a fixed property of the machine: it is roughly 1% when you scan twice in an afternoon, and meaningfully worse across days and months, because your own body water shifts between visits.
These figures come from specific populations, mostly healthy or athletic adults. Agreement tends to be poorer at the extremes of body composition, and a study of older women using an InBody 720 found wide limits of agreement at the individual level even where the group average was close. [6]
Why InBody reads lower than DEXA
The largest head-to-head comparison available measured 795 active-duty US Marines on both an InBody 770 and a GE Lunar iDXA. InBody predicted DXA body fat percentage with a mean absolute error of 3.9 percentage points, but with an r² of 0.96, meaning it followed the pattern across people very closely. When the authors added a correction of roughly 3 percentage points to every InBody reading, the two lined up. [2]
A standing impedance device tracked DXA body fat closely in pattern while sitting consistently below it, an offset a fixed correction largely removed.
— Adapted from the InBody 770 and iDXA comparison in 795 US Marines
A consistent offset behaves very differently from random noise. Random error makes every reading unreliable. A systematic offset leaves each reading shifted by a similar amount, so the difference between two readings on the same device still tells you something real. That is why an InBody number can be trustworthy for tracking and still not match a DXA printout.
The direction is not universal. The study of older women found a slight overestimate rather than an underestimate, with a spread of roughly seven percentage points either side at the individual level. [6] Population, hydration, and the device's equations all play a part, which is why the useful rule is not "add three points" but "compare like with like."
Which one to use for tracking change
Here is the practical version, using the numbers above to read a hypothetical twelve-week training block.
| Scenario | What it likely means |
|---|---|
| InBody drops 0.5 points, same device and conditions | Within the measurement error of the device; too small to call a change |
| InBody drops 3 points, same device and conditions | Comfortably larger than measurement error; a real shift |
| DEXA fat mass drops 400 g across two visits | Close to the day-to-day precision error; treat cautiously |
| DEXA fat mass drops 2 kg across two visits | Well beyond precision error; a real shift |
| InBody says 18%, DEXA says 21% | The expected offset between methods, not a change in your body |
Reading a change over twelve weeks. Figures are drawn from the precision studies cited above and are illustrative of how to interpret a result, not thresholds for any individual.
The last row is the one that catches people out. Someone who scans on an InBody in January and a DEXA in April can conclude they have gained three points of body fat when nothing has happened at all. Switching devices mid-programme destroys the comparison.
This is why consistency beats precision for anyone following progress. At Different Health, body composition is measured on the same in-lab InBody at every visit, alongside skeletal muscle mass and visceral fat, so the comparison between visits stays valid rather than being reset by a change of equipment.
When a DEXA is worth adding
There are three situations where the extra accuracy earns its cost and its few minutes of low-dose radiation.
The first is bone. DEXA measures bone mineral density and InBody does not, so if bone health is the question, impedance cannot answer it. The second is a regional breakdown of fat: DEXA maps fat and lean tissue limb by limb, while InBody reports lean mass by segment without the same fat detail. The third is an accurate baseline. If you want your starting number to be as close to true as possible, rather than internally consistent, DEXA is the better single measurement.
One caveat that applies to DEXA as much as to impedance: results are not interchangeable between manufacturers. Reference values built on GE systems are explicitly not transferable to other DXA machines, [7] so a scan at one facility will not line up cleanly with a scan at another. Radiation exposure is minimal either way, in the range of 0.001 to 0.01 millisieverts. [8] As with any X-ray-based test, DEXA is generally avoided in pregnancy. This is general information rather than personal medical advice; anything specific to your situation belongs with your own clinician.
Using both without confusing yourself
Plenty of people end up with readings from both, and the two can work together provided you keep them in separate lanes. Use DEXA for an accurate baseline and for anything involving bone or regional detail. Use InBody for the frequent check-ins in between. Compare each result only against previous results from the same method, and expect a standing gap of a few percentage points between the two.
What neither method does is tell you what to do next. A body fat percentage is a measurement, and its value depends entirely on being read in context: against your own history, against healthy ranges for your age and sex, and alongside the rest of your metabolic and strength picture. Different Health treats the number that way, with a team of MDs and PhDs reading body composition next to your cardiorespiratory and strength data and turning the whole picture into a nutrition and training plan you can retest against.
If you are still deciding where to get scanned, our guide to DEXA scans in NYC covers what providers charge and how to choose between them.
Key takeaways
- DEXA is more accurate, InBody more repeatable. One gives the better single number, the other the better habit.
- Expect a gap of about 3 points. In 795 Marines, InBody sat systematically below DXA and a fixed correction closed it.
- A systematic offset is not noise. Because the shift is consistent, InBody still shows real change between visits.
- InBody repeats to within a point. Standard error of measurement was 0.77–0.99 percentage points across three models.
- DEXA's precision depends on timing. Under ~1% for same-day repeats, and worse across days and months.
- Never switch devices mid-programme. Comparing an InBody reading to a DEXA reading invents a change that didn't happen.
Frequently asked questions
Is InBody as accurate as DEXA?
Not for a single absolute number, but it is close on the thing most people care about. In a study of 795 active-duty US Marines, the InBody 770 tracked DXA body fat percentage very closely in pattern, with an r-squared of 0.96, while carrying a mean absolute error of 3.9 percentage points. The gap is largely systematic rather than random, which means InBody is reliable for following your own trend even where it does not match a DXA figure exactly.
Why is my InBody body fat different from my DEXA result?
Because the two methods measure different things and one is calibrated against the other. DEXA separates fat, lean tissue, and bone using two X-ray beams. InBody estimates body composition from how a small electrical current travels through the body, then converts that into fat and lean mass using the manufacturer's equations. In the Marines study, applying a correction of about 3 percentage points to the InBody figures brought them into line with DXA, which is the size of gap most people should expect.
Which is better for tracking progress, InBody or DEXA?
For frequent tracking, InBody has a practical edge. Testing found the InBody 230, 720, and 770 highly repeatable, with intraclass correlations at or above 0.98 for body fat percentage and a standard error of measurement between 0.77 and 0.99 percentage points. It also takes under a minute and involves no radiation, so it can be repeated often. DEXA gives the more accurate single snapshot. Whichever you choose, staying on one device matters more than the choice itself.
How much does a DEXA scan vary between scans?
Less than most methods, but not zero. Repeat scans on the same day with repositioning show precision errors under about 1 percent for total body masses. Across longer intervals the variation grows: one study of resistance-trained athletes found consecutive-day precision error for fat mass was nearly double the same-day figure, at 1,261 grams versus 660 grams. Visceral fat is the least stable measure of all, which is why small shifts in it should be read cautiously.
Does InBody underestimate body fat?
In the largest comparison available, yes. Among 795 US Marines measured on both an InBody 770 and a GE Lunar iDXA, InBody readings sat systematically below DXA, and adding about 3 percentage points brought them into agreement. Findings differ by population: a study of older women using an InBody 720 found a small overestimate on average, with wide limits of agreement at the individual level. The practical takeaway is that an InBody number is not interchangeable with a DXA number in either direction.
Should I get a DEXA scan if I already use InBody?
It depends on what you need the number for. If you are following your own trend over months on the same device, InBody answers that well on its own. A DEXA is worth adding when you want the most accurate single reading, when you need bone mineral density, or when you want a regional breakdown of fat and lean mass. Many people use both: DEXA for an accurate baseline, InBody for the frequent check-ins in between.
References
- Nickerson BS, et al. Reliability and agreement of various bioelectrical impedance analyzers as compared to dual-energy X-ray absorptiometry in healthy men and women. Journal of Clinical Densitometry. (ICC ≥0.98 and SEM 0.77–0.99% for body fat percentage across three analyzer models.)
- High precision but systematic offset in a standing bioelectrical impedance analysis compared with dual-energy X-ray absorptiometry. (Standing impedance vs Lunar scanner in 480 men and 315 women; mean absolute error 3.9 percentage points for body fat, r² = 0.96, corrected by adding ~3 points.)
- Precision of Lunar dual-energy X-ray absorptiometry in measuring body composition among colorectal cancer patients and healthy subjects. Clinical Nutrition ESPEN. (Total-region precision errors of 0.49–1.01% and 0.40–0.88%.)
- Same-Day Vs Consecutive-Day Precision Error of Dual-Energy X-Ray Absorptiometry for Interpreting Body Composition Change in Resistance-Trained Athletes. Journal of Clinical Densitometry. (Consecutive-day fat-mass precision error 1,261 g vs 660 g same-day.)
- Reliability of Compartmental Body Composition Measures in Weight-Stable Adults Using GE absorptiometry: Implications for Research and Practice. (CV 0.8–5.9% over three months versus under 2% same-day.)
- Comparison of body fat percentage assessments by bioelectrical impedance analysis, anthropometrical prediction equations, and dual-energy X-ray absorptiometry in older women. Frontiers in Nutrition. (Impedance analyzer concordance 0.805, slight overestimate, limits of agreement −7.03 to +6.22.)
- Imboden MT, et al. Reference standards for body fat measures using GE dual energy x-ray absorptiometry in Caucasian adults. PLOS ONE. 2017;12(4):e0175110. (Values not interchangeable across absorptiometry manufacturers.)
- Australian Radiation Protection and Nuclear Safety Agency (ARPANSA). Statement on the use of Dual-Energy X-ray Absorptiometry. (Radiation dose range 0.001–0.01 mSv.)