Training & Nutrition · Sep 2, 2026

Your DEXA says body fat dropped but weight went up — what actually happened

Fat mass down and scale weight up is usually a good scan, not a contradiction. Here is how to tell muscle from water, what is inside the measurement error, and what to change in the plan.

Short answer

Why did my DEXA show less body fat but my weight went up? Because the two measurements are not counting the same thing. A scale reports total mass, including water, glycogen and gut contents, while a DEXA separates that mass into fat, lean soft tissue and bone. Fat mass falling while total weight rises usually means something that is not fat went up, most often stored carbohydrate and the water held with it, and less often new muscle.

Is the extra lean mass on my DEXA muscle or water? Timescale is the best guide. Muscle is gained slowly, in fractions of a pound per month for anyone past their first year of training, so a several-pound lean-mass jump over a couple of months is more likely to be glycogen and water. Each gram of glycogen is stored with roughly three grams of water, so returning to normal carbohydrate intake or simply training harder can add real weight that is not tissue.

How much of a DEXA change is measurement error? Repeat scans of the same person typically vary by about one to two percent on body fat percentage, and readings for a single arm or leg vary more than whole-body ones. A change smaller than roughly a percentage point of body fat, or about a pound of fat mass, sits inside that error band and should not be treated as a result.

How often should you get a DEXA scan? Ten to sixteen weeks apart is usually far enough for a real change to clear the measurement error. Scanning every four weeks mostly returns noise, because the error band is close to the size of the change a month can produce.

Should you add cardio if your weight went up after a scan? Usually not, if fat mass fell. Adding conditioning to force the scale number down costs recovery and works against the lean mass the scan just credited you with. Food intake is the more direct lever when fat mass has genuinely stalled across two scans.

The two numbers are not measuring the same thing

A DEXA scan and a bathroom scale answer different questions, and the confusion starts there.

The scale reports one number: everything in your body, all at once. Fat, muscle, bone, organs, the water in your bloodstream and muscles, the food in your gut, and whatever you drank in the last hour. It cannot separate any of it.

A DEXA separates that mass into fat, lean soft tissue, and bone mineral. So when your scan says fat mass dropped three pounds while the scale says you are two pounds heavier, both readings can be correct at once. Five pounds of something that is not fat went up while fat went down.

The useful question is not which number to believe. It is which of the four realistic explanations you are looking at, because two of them mean the plan is working and two mean it needs a change.

The four things that actually cause it

Work down this list in order. The first two are far more common than people expect.

Notice that three of the four are about water and measurement, not about tissue. That is the single most useful thing to know before you change anything.

Glycogen and the water stored with it. Muscle stores carbohydrate as glycogen, and each gram of glycogen is stored alongside roughly three grams of water. Adding carbohydrate back after a low-carb stretch, or simply training harder in the weeks before the scan, can add several pounds that a DEXA counts as lean tissue. It is real weight and it is not muscle.
Everyday fluid variation. Sodium intake, sleep, a hard session two days before, travel, alcohol, and the menstrual cycle all move total body water by a few pounds in either direction. If your two scans were not taken under similar conditions, some of the difference is this.
Actual muscle gain. This is the answer everyone hopes for and it is the slowest of the four. Someone new to lifting can add muscle quickly enough to offset fat loss on the scale. Someone several years in is generally working in fractions of a pound per month, so a four-pound lean-mass jump over eight weeks is much more likely to be the first two explanations than pure tissue.
Scan noise. Two scans of the same person on the same day do not produce identical numbers. More on that below, because it is the explanation people skip.

How much of this is just scan noise?

Every measurement has an error band, and a DEXA is no exception. Repeat scans on the same machine typically vary by around one to two percent on body fat percentage, and regional readings for a single arm or leg vary more than whole-body ones do.

The practical consequence: a change smaller than roughly a percentage point of body fat, or a pound or so of fat mass, is inside the noise. It is not a result. It is the machine.

You can shrink that band considerably by controlling the conditions:

If you cannot say yes to most of those for both scans, treat the comparison as directional rather than precise.

Use the same machine every time. Cross-machine comparisons are not reliable, even between two units of the same model.
Scan in the morning, fasted, before training and before large fluid intake.
Keep the days before a scan ordinary. Do not scan the morning after a long run, a heavy leg session, a flight, or an unusually salty meal.
Keep carbohydrate intake roughly typical in the days before. A depleted scan compared against a loaded one exaggerates every lean-mass difference.

Read the report in this order

Most people open a DEXA report and go straight to body fat percentage, which is the number most affected by everything above. A more stable reading comes from taking them in this order.

Fat mass down with weight up is a good scan. Say that out loud, because the scale spent the last eight weeks telling you the opposite.

Fat mass in pounds or kilograms. This is the number you are actually trying to move, and unlike a percentage it does not shift when lean mass changes.
Lean soft tissue in pounds. Compare it against how long you have been training and how much your carbohydrate intake has changed. A jump larger than a plausible rate of muscle gain is a water story.
Body fat percentage. Useful, but it is a ratio of the first two, so it moves when either one does. It can fall while fat mass is flat.
Regional breakdown. Directionally interesting, individually noisy. Do not build a training decision on one limb.

What to change in training, and what to leave alone

If fat mass fell and lean mass held or rose, the honest answer is that almost nothing needs to change. The block did what a block is supposed to do. The most common mistake here is adding cardio to chase the scale number, which attacks the one part of the result that was already going the right way.

If fat mass was flat and lean mass rose, the weight gain is intake, water, or both, and the lever is food rather than training.

If fat mass rose while lean mass held, that is a straightforward intake question, and adding sessions is the least efficient way to answer it.

Keep lifting the way you were. Strength work is what protected the lean mass that made this scan look the way it does.
Do not add conditioning to force the scale down. If fat mass is falling at a reasonable rate, more cardio buys you very little and costs recovery you were using to progress.
Keep progressing load or reps on the main lifts. A plan that keeps getting slightly harder is what keeps lean mass where it is.
Change one variable at a time. If you adjust both training volume and intake before the next scan, you will not know which one moved the number.

What to change in food

Nutrition is the faster lever here, and it needs less adjustment than most people apply after a scan.

Hold protein steady. Protein intake is what makes fat loss show up as fat loss rather than as lost lean mass on the next scan.
If fat mass is falling at a rate you are happy with, change nothing at all. A working intake does not need a smaller number attached to it.
If fat mass is flat over eight or more weeks, adjust total intake modestly rather than cutting hard. A large cut tends to take lean tissue with it, which is visible on the following scan.
Do not read the day-to-day scale as feedback on any of this. Judge intake on a three to four week rolling average, and on the next scan.

How often to rescan

More scans do not produce more information. If the measurement error is around a percentage point and a realistic rate of change is one to two pounds of fat per month, scanning every four weeks means most of what you see is noise.

Ten to sixteen weeks apart is far enough for a real change to clear the error band. That interval also lines up with the length of a training block, which is the thing you would actually adjust on the strength of a scan.

What CoachArc does with a scan like this

A scan is only worth the plan change it produces, and that is the part almost nobody writes about.

CoachArc reads a fat-down, weight-up scan as confirmation rather than a problem. It holds the training structure that produced it, keeps protein where it is, leaves conditioning alone, and continues progressing the lifts. When fat mass has genuinely stalled across two scans, it moves the intake lever before it adds sessions, because the sessions are what protected the lean mass in the first place.

The scan does not change your goal. It tells you whether the current plan is still the right way to get there, and most of the time the answer after a scan like this one is yes.

A DEXA measures body composition. It does not diagnose anything, and bone density findings or anything else on the report that looks clinical belongs with your own healthcare provider, not with a training plan.

How CoachArc uses this idea

CoachArc is built around practical, adaptive coaching: what to do today, what changed, and how to keep momentum without turning a missed day or imperfect meal into a restart.

It connects today's plan, food guidance, progress, and Hey Coach so you review a recommendation and apply it only when it makes sense. It reads the health data you choose to share — body-composition scans, blood panels, and wearable recovery — and proposes the change it wants to make.

See everything CoachArc does, how it compares to trackers and generic AI, or what it costs.