Why did my body fat percentage jump between two DEXA scans?

Twenty-one resistance-trained athletes were scanned twice in one sitting, then again the next day. Within a single sitting the precision error for whole body fat mass was 660 g. Across two days it was 1261 g, and lean mass went from 617 g to 2083 g, more than triple (see [1]). Nobody gained 1.4 kg of tissue overnight. The scanner measured the same people under different conditions and reported the difference as body composition.

That matters because of how a real change gets defined. The International Society for Clinical Densitometry sets the smallest trustworthy change at 2.77 times the precision error, the 95 percent confidence threshold (see [2]). Apply that to the consecutive-day figure above and fat mass has to move roughly 3.5 kg in that population before the result is separable from noise. On a 75 kg body that is close to five percentage points of body fat. Most of what people read off a second scan a month later sits well inside that band, which is worth knowing before you interpret the number at all, and our body fat percentage guide covers what the reading is actually composed of.

The error is not fixed, though. Twenty recreationally active adults recorded every food, fluid and activity input in the 24 hours before their first scan and replicated it before the next one. Fat mass precision error came in at 479 g on the same day and 626 g across days, with no significant difference between them (see [3]). Standardizing the preceding day removed most of the biological gap. A separate trial in thirty-two athletic males compared fasted with non-fasted presentation and found that DXA and air displacement plethysmography stayed acceptable as long as acute food and fluid intake stayed under 500 g, while models estimating total body water were far more sensitive to a non-standardized state (see [4]).

So the protocol is fixed conditions, not more frequent scans: same machine, first thing in the morning, fasted and rested, with the previous 24 hours of food, fluid and training deliberately repeated. Log those conditions next to the result in Wellness Project under /body, and compare quarterly rather than monthly, because the change has to outrun the error band before it carries information. A scan does not measure your body. It measures your body plus the previous day, and only one of those is the thing you are tracking.