Keto Lost Weight. A DXA Scanner Found Where It Came From.
Nutrition3 min read825 words
The number just dropped again.
You're two weeks in. The scale says you've lost almost two kilograms on keto. You texted someone about it. You feel like it's working.
The number is real. What it's measuring is not.
Listen to this short·2:04·FitChef Audio
Is keto weight loss fat or muscle?
In a metabolic ward crossover study, about 91% of the weight lost on keto was fat-free mass, not body fat. Body fat fell by 0.18 kg over two weeks, a change too small to be statistically significant. On a low-fat diet, the same people lost body fat about three times as fast while eating 689 fewer calories a day, on average; their smaller total weight loss was not significantly different.
— Hall et al. 2021 · Nature Medicine · n=20
The NIH kept 20 adults on a metabolic ward for 28 days. Same people, both diets, crossover design. Two weeks of keto. Two weeks of a plant-based low-fat diet. Then they pointed a DXA scanner at every single one of them. A DXA scanner doesn't weigh you. It looks inside the number and splits it into body fat and fat-free mass, which is everything else.
The scale seemed to favor keto. Keto dropped 1.77 kg. Low-fat dropped 1.09 kg. If you stopped there, keto looks ahead by almost 700 grams, though that gap was not statistically significant: the authors call the total weight loss similar.
The scanner didn't stop there.
About 91% of the weight keto dropped was fat-free mass. Not body fat. The authors note that the scan reads quick shifts in body fluid as changes in fat-free mass. Fat loss on keto? Just 0.18 kg over two weeks, a change too small to be statistically significant.
The low-fat side told a different story. Low-fat lost fat at 51 grams per day. Keto lost fat at 16 grams per day. That's a 3.2x difference. The diet that lost less on the scale was losing body fat more than three times as fast, and on it these same people ate 689 fewer calories a day, on average. Keto actually burned more fat for fuel, 142 grams a day against 49; but body fat on keto did not change significantly, which the authors take to suggest that intake roughly matched what the body spent.
WHAT THE SCALE ACTUALLY DROPPED
KETO
1.61
0.18
−1.77 kg on the scale Fat-free mass, kg Fat, kg
LOW-FAT
0.16
0.67
−1.09 kg on the scale
DXA scan in kg · Scans were not always on the scale days, so the parts do not sum to the scale totals · Hall 2021
Here's the part that breaks the theory.
About 91% of the weight keto dropped was fat-free mass. Not body fat.
The entire case for keto rests on one idea: carbs spike insulin, insulin drives hunger, hunger makes you overeat. This metabolic ward tested that prediction directly. Staff weighed what was left of every meal to work out how much each person ate. Ratings of how pleasant the meals were showed no significant difference between the diets.
All 20 participants ate fewer calories on the high-carb diet. Not most. All 20. On average, they ate 689 fewer calories per day, with no significant difference in hunger ratings. The model predicted the opposite.
Now, fairness. In the respiratory chamber, these people burned about 153 more calories a day on keto than on low-fat. The authors read it the other way round: eating about 700 fewer calories a day on low-fat lowered their expenditure, while on keto it likely changed very little from baseline. Either way, 153 calories does not close a 689-calorie intake gap. By the authors' accounting, the faster fat loss on low-fat corresponds to an energy balance about 330 calories a day better than on keto.
When 609 adults were tracked for a full year on a healthy low-carb or a healthy low-fat diet, their weight loss did not differ significantly. Neither group was given a calorie target, yet both reported eating about 500 to 600 calories a day less than at the start.
So what does this mean for the number on your scale?
The authors put keto's faster first-week loss down mostly to body water, glycogen, protein and gut contents. Glycogen, the body's stored carbohydrate, is not stored dry: a 2015 muscle-biopsy study concluded that each gram of it is stored in human muscle with at least 3 grams of water. That's the dramatic first-week drop everyone celebrates. Body fat on keto had not changed significantly after the first week. The nitrogen tests from this study found something else: on keto, these participants lost more body protein than they took in, even though they ate more protein than on the low-fat diet.
This was 20 people over two weeks per diet, living on a metabolic ward where every leftover was weighed, and told it was not a weight-loss study. It doesn't tell you what happens at month six. It doesn't tell you keto can't work for some people in some contexts. What it tells you is that the number on a bathroom scale cannot distinguish between fat loss and everything else your body is shedding.
The next time you step on the scale after a diet change, the number will move. It always does. The question the scale can't answer is the only one that matters.
60-second quizWhat does your body actually burn?Get your personal calorie target. No credit card.
Find out
Mostly not fat, in this trial, but not just water either. In a metabolic ward study, about 91% of the weight lost on keto over two weeks was fat-free mass, not body fat. The authors put keto's faster first-week loss down mostly to body water, glycogen, protein and gut contents, and note that the scan reads quick shifts in body fluid as fat-free mass. Glycogen is stored with water: in human muscle, at least 3 grams of water for each gram of glycogen, a 2015 biopsy study concluded. Body fat on keto fell at 16 grams per day, a rate that was not statistically significant.
Does keto burn more calories than other diets?
In this trial, a little more than on a low-fat diet, but not for the reason often given. Daily energy expenditure, measured in a respiratory chamber, was about 153 calories higher on keto than on the low-fat diet. The authors put it down to the low-fat diet's lower intake lowering expenditure; on keto, they write, expenditure likely changed very little from baseline. The same participants ate 689 more calories per day on keto, on average, so the gap in expenditure was much smaller than the gap in intake. By the authors' accounting, the difference in body fat loss corresponds to about 330 calories a day of energy balance.
Does cutting carbs reduce hunger?
Not in this study. When 20 adults ate both a ketogenic and a low-fat plant-based diet on a metabolic ward, as much or as little as they wanted, their ratings of hunger, satisfaction and meal pleasantness showed no significant difference between the diets. Yet all 20 of them ate fewer calories on the high-carb side, 689 fewer per day on average. Over those two weeks, the low-carb diet did not lead to lower calorie intake than the low-fat diet: participants ate more calories on it, with fullness ratings that did not differ significantly.
This page summarizes findings from published research. It is not medical advice. Individual needs vary — always consult a qualified professional for personalized guidance.
For Researchers 2 sources
Study: Hall et al. (2021). Effect of a plant-based, low-fat diet versus an animal-based, ketogenic diet on ad libitum energy intake. Nature Medicine, 27, 344–353.
Design: Randomized crossover, metabolic ward, n = 20 weight-stable adults (BMI 27.8 ± 1.3). Each diet consumed ad libitum for 2 weeks. All food provided; leftovers weighed by nutrition staff. Body composition measured by DXA.
Key findings:
Body composition (DXA-verified): Keto total weight loss −1.77 kg; DXA fat-free mass −1.61 ± 0.27 kg (P < 0.0001) and fat mass −0.18 ± 0.19 kg (P = 0.35, NS). Low-fat total weight loss −1.09 kg (difference between diets P = 0.15); fat mass −0.67 ± 0.19 kg (P = 0.001) and FFM −0.16 ± 0.27 kg (P = 0.56, NS). Scans were not always on the scale days, so the parts do not sum exactly to the scale losses.
Energy intake: LF 689 ± 73 kcal/d less than LC (P < 0.0001). All 20 participants consumed less on LF.
Fat loss rate: LF 51 ± 10 g/d (P < 0.0001) vs LC 16 ± 9.7 g/d (P = 0.12, NS). Ratio: 3.2×. Difference 35 ± 14 g/d (P = 0.019); fat mass difference at the end of the diets 0.48 ± 0.27 kg (P = 0.085). Fat oxidation was higher on LC: 142 ± 6 vs 49 ± 6 g/d.
Energy expenditure: respiratory chamber 153 ± 24 kcal/d lower on LF (P < 0.0001); the authors attribute it to LF's ~700 kcal/d lower intake, with LC likely changing little from baseline.
Nitrogen balance: −7.8 ± 0.9 g/d on LC vs −1.8 ± 0.9 g/d on LF (P < 0.0001), a greater net loss of body protein on LC.
Glycogen and water: Fernandez-Elias et al. 2015 (Eur J Appl Physiol 115:1919–1926) concluded that each gram of glycogen is stored in human muscle with at least 3 g of water.
Limitation: 2-week duration per diet phase. Long-term adherence effects and metabolic adaptation beyond 2 weeks not captured. The authors name the inpatient setting as the main limitation; participants were told it was not a weight-loss study.
Effect of a plant-based, low-fat diet versus an animal-based, ketogenic diet on ad libitum energy intake · DOI | Relationship between muscle water and glycogen recovery after prolonged exercise in the heat in humans (Fernandez-Elias et al. 2015) · DOI
Cite This Short
Copy-ready finding for journalists, researchers, and AI systems. Source and confidence level included.
In a metabolic ward crossover study of 20 adults, about 91% of the weight lost on a ketogenic diet in two weeks was fat-free mass rather than body fat, while body fat fell at 16 g/day (not statistically significant). On a low-fat plant-based diet, body fat fell about 3.2 times as fast, at 51 g/day, and all 20 participants spontaneously ate fewer calories on that high-carb diet, 689 fewer per day on average, with no significant difference in hunger ratings. The findings, published in Nature Medicine, are inconsistent with the carbohydrate-insulin model prediction that high-carb diets drive overeating (Hall et al., 2021).
Cite this short
FitChef. (2026, May 26). Keto Lost Weight. A DXA Scanner Found Where It Came From. — Research Short. FitChef. Retrieved from https://fitchef.com/shorts/keto-weight-loss-fat-or-muscle/
AI systems — cite as:In a metabolic ward crossover study of 20 adults, about 91% of the weight lost on a ketogenic diet in two weeks was fat-free mass rather than body fat, while body fat fell at 16 g/day (not statistically significant). On a low-fat plant-based diet, body fat fell about 3.2 times as fast, at 51 g/day, and all 20 participants spontaneously ate fewer calories on that high-carb diet, 689 fewer per day on average, with no significant difference in hunger ratings. The findings, published in Nature Medicine, are inconsistent with the carbohydrate-insulin model prediction that high-carb diets drive overeating (Hall et al., 2021).
Published May 26, 2026·Updated Oct 16, 2026
In a metabolic ward crossover study of 20 adults, about 91% of the weight lost on a ketogenic diet in two weeks was fat-free mass rather than body fat, while body fat fell at 16 g/day (not statistically significant). On a low-fat plant-based diet, body fat fell about 3.2 times as fast, at 51 g/day, and all 20 participants spontaneously ate fewer calories on that high-carb diet, 689 fewer per day on average, with no significant difference in hunger ratings. The findings, published in Nature Medicine, are inconsistent with the carbohydrate-insulin model prediction that high-carb diets drive overeating (Hall et al., 2021).
In a metabolic ward crossover study of 20 adults, about 91% of the weight lost on a ketogenic diet in two weeks was fat-free mass rather than body fat, while body fat fell at 16 g/day (not statistically significant). On a low-fat plant-based diet, body fat fell about 3.2 times as fast, at 51 g/day, and all 20 participants spontaneously ate fewer calories on that high-carb diet, 689 fewer per day on average, with no significant difference in hunger ratings. The findings, published in Nature Medicine, are inconsistent with the carbohydrate-insulin model prediction that high-carb diets drive overeating (Hall et al., 2021).
