A 164-person trial found low-carb burns 278 extra calories a day after weight loss. Then an independent team reanalyzed the raw data — and the number kept getting smaller.
“The metabolic boost was 3.6 times bigger for some people than others — 478 extra calories per day for high insulin responders versus 131 for low. Same diet, same controlled meals, vastly different bodies.”
The number is real.
In 2018, a research team at Boston Children's Hospital published what remains the largest controlled-feeding trial on this question. It tested whether staying low-carb after weight loss gives your body a metabolic edge.
They took 164 adults who had just lost 10-12% of their body weight, randomly assigned them to diets with 20%, 40%, or 60% of calories from carbohydrates, and prepared every single meal for twenty weeks. Same total calories. Same protein. Only the carb-to-fat ratio changed.
The low-carb group burned 278 extra calories per day compared to the high-carb group — roughly 10% of a full day's food, gone without eating less or moving more. Their bodies simply used more fuel while maintaining the same weight, doing the same activities, sleeping in the same beds.
If you heard that number in a David Ludwig lecture or a low-carb podcast, you heard it right. The paper was published in the BMJ. The study population — mostly women, average age 38, overweight to obese before their initial weight loss — ate meals weighed to within 5 grams of target.
Blood tests confirmed the diets were genuinely different. Lipid panels, insulin markers, and a carb-specific blood test all separated cleanly between groups.
The finding is significant. The trial is serious. And everything that follows makes that number more interesting, not less.
The largest controlled-feeding trial on low-carb metabolism found an extra 278 calories burned per day — then independent teams reanalyzed the same data and the number started shrinking.
- The study fed 164 adults every meal for 20 weeks after weight loss and found the low-carb group burned 209-278 extra calories per day compared to the high-carb group.
- The metabolic boost depended on the person's biology — people with the highest insulin responses got 3.6 times more benefit than those with the lowest.
- Independent reanalysis showed the number shrinks to as low as 46 calories per day after correcting for how the measurement tool handles different diets.
- The hunger hormone ghrelin declined more on the low-carb diet — a secondary finding suggesting low-carb may help with appetite during weight maintenance.
- Even the lead critic acknowledged a real metabolic effect might exist — at 50 to 140 calories per day, not the 278 from the headlines.
The Metabolic Edge Depends on Who You Are
Buried in the same paper is a finding the clips almost never mention.
Ebbeling's team measured how much insulin each person's body released 30 minutes after drinking a sugar drink — a standard test of how strongly the body reacts to sugar. Then they split the results by thirds: high responders, medium, and low.
The metabolic boost wasn't one number — it was three.
People in the highest insulin third burned 478 extra calories per day on the low-carb diet. People in the lowest third burned 131. Same study, same protocol, same controlled meals — a 3.6-fold difference in metabolic response, sorted by a single blood marker measured before the trial began.
The study didn't test why insulin response predicted such a dramatic split — only that it did. But the implication is hard to ignore. Whatever metabolic edge low-carb diets might provide during weight maintenance, the size of that edge appears to depend less on the diet and more on the biology of the person eating it.
Which raises an uncomfortable question. If the effect varies this much from person to person — how confident should anyone be in the 278-calorie average?
When the Number Started Shrinking
Between the BMJ paper and the next viral clip, something happened that the clips never cover.
Ebbeling's team measured calorie burning with a lab method that tracks the CO₂ your body produces. But turning that CO₂ reading into a calorie number requires a formula — and that formula has to guess what mix of fuels your body is burning. Fat, carbs, or protein. That guess changes depending on the diet.
In 2019, Kevin Hall published a paper looking at exactly this assumption. Hall, an NIH researcher, had been debating David Ludwig about the carb-insulin model for years. [1]
He used data from a tightly controlled feeding study with the same measurement tool. He showed what happens when you correct the fuel-mix estimate for what actually occurs during a low-carb diet. The apparent metabolic advantage dropped: 209 became 139 calories per day. The statistical significance disappeared. [1]
Then he removed people whose energy accounts didn't add up. These were people whose measured calorie burning topped their food intake by over 600 calories a day — despite stable body weight. The number dropped again: 139 became 46. [1]
But Hall didn't stop there. He downloaded Ebbeling's raw data — all 164 people — from a public repository and looked at what the CO₂ sensors had actually recorded. Not the converted calorie estimate. The raw gas people breathed out. [2]
No significant difference between the diet groups. Not under the original analysis. Not under the revised analysis. Not under any model Hall tried. The entire calorie gap — all of it — lived inside the formula's fuel-mix assumption, not in anything the instruments actually picked up. [2]
209 → 139 → 46 → 0. Four numbers, each produced by stripping one more layer of mathematical assumption from the same type of measurement. The distance between the published result and the tightest correction is roughly the energy in one chocolate chip cookie.
The Plan That Changed After the Data
Hall's review of Ebbeling's data turned up something beyond the measurement debate. [2]
Before the trial began, Ebbeling's team filed a written plan — locking in exactly how they would crunch the numbers before seeing them. The original plan compared calorie burning during the test phase to a baseline measured before people lost weight.
Under that plan, after all 164 people had finished the trial, the diet effect on calorie burning was indistinguishable from chance — a 43% probability of seeing that result even if the diet changed nothing at all. [2]
The published result — the one that launched a thousand clips — came from a revised analysis plan that anchored the comparison to a different time point. The revision was filed after all people had finished the study. The paper did not report the results of the original plan. [2]
Changing an analysis plan after collecting data isn't always wrong — researchers sometimes have good reasons. But when one plan gives you a 43% chance of being noise and the other gives you a near-zero chance, that matters. It changes how much weight a finding can carry.
“Even the lead critic of the finding conceded a real metabolic effect might exist — at 50 to 140 calories per day, not 278.”
What Both Sides Actually Agree On
If this were a simple debunking — study published, study destroyed, everyone moves on — the story would end here. It doesn't.
Ebbeling's team published a formal response defending their methods and their decision to revise the analysis plan. And Hall, in his own paper, said something the most confident skeptics skip. He "cannot definitively exclude the possibility of a real increase in energy expenditure, especially at an effect size of ~50-140 kcal/d." [1]
That range — 50 to 140 calories per day — is a long way from the 278 in the viral clips. But it's not zero. It might be real. It might depend on insulin response. And the measurement debate isn't settled.
A separate team tested the same question under even tighter control — a metabolic ward where every morsel was tracked and no one could sneak outside food. They found roughly 57 extra calories per day on a ketogenic diet. The researchers called it "near the limits of detection with the use of state-of-the-art technology." [3]
The truth, as far as anyone can pin it down today, probably lives somewhere between 278 and zero. Smaller than the clips promise. Possibly real. And almost certainly not the same size for everyone. How a banana-sized daily edge ranks against the variables that dwarf it is what thirty-three studies across nine questions reveal.
What This Fight Actually Gives You
If you came to this page from a Ludwig clip expecting the 278 to be settled science, you now have something the clip didn't offer: the full fight.
Three independent challenges, each using the study's own data or the same measurement tool. A changed analysis plan. A CO₂ reading that showed nothing.
And on the other side — a well-designed trial, a significant finding in a major journal. An insulin subgroup result that no critic has fully explained away. An independent metabolic ward study that found a smaller but non-zero effect.
209 is one answer. 46 is another. Zero is a third. 50-140 is what even the critics concede. The question isn't which number is right. The question is which assumptions you accept — and whether your own biology puts you closer to the 478 end of the spectrum or the 131 end.
The number from the viral clip was 278 extra calories per day. After adjusting for the factors that inflate that measurement, the range that even the critics acknowledge as possibly real lands at 50 to 140 calories per day — roughly a handful of almonds, not a free workout.
When the same data was re-run with controls for the participants' body composition, age, sex, and starting metabolism, the advantage dropped to 185 calories per day. Still meaningful. Still smaller than the clip promised.
The study also found a sliding scale: for every 10% you reduce carbohydrates, energy expenditure went up by about 52 calories per day. That means partial carb reduction — not full keto — might carry a proportional fraction of whatever real effect exists.
If you restructured your maintenance meals around a 200-300 calorie metabolic bonus, the honest calibration is that the bonus is probably real but two to four times smaller than the number you heard — and how much of it applies to you may depend on your insulin response.
What other research found
What this means for you
The metabolic boost wasn't evenly distributed. People whose bodies released the most insulin after a glucose drink saw the largest energy expenditure increase on the low-carb diet — more than three times bigger than what people in the lowest insulin group measured.
Even when the researchers counted everyone who started — including those who didn't follow the diet perfectly — the high-insulin group still showed a meaningful difference of 308 extra calories per day.
The same measurement debate applies to this subgroup number — it inherits the same assumptions about the tool used to measure energy. But no critic has fully explained away why insulin response predicted such a dramatic split in the first place.
The metabolic boost for people with the lowest insulin responses came in at 131 extra calories per day among participants who stuck closely to the diet. That number is already inside the range that critics say might just be measurement noise.
If the real effect is closer to 50-140 calories per day — which even the lead critic considers possible — then for low-insulin individuals, the metabolic edge of staying low-carb during maintenance may be too small to build a strategy around.
That doesn't mean low-carb has no benefits for this group. It means the metabolic-rate argument specifically — burning extra calories just by changing macros — carries less weight here than other reasons someone might choose lower carbohydrates.
Before you change anything
This study tested 164 overweight and obese adults — people with a BMI of 25 or higher who had already lost about 10-12% of their body weight before the diet phase began.
The average participant was around 38 years old, weighed roughly 90 kilograms before their initial weight loss, and was maintaining that loss — not actively trying to lose more. About 70% were women and 78% were white, all recruited from a single university campus in Massachusetts.
Every meal was prepared and delivered by the study team. The participants didn't choose their food, shop for groceries, or cook. If you're making your own meals — which is nearly everyone reading this — the controlled-feeding results may not translate directly to your kitchen.
The biggest gap between this study and your life is the controlled feeding. Every meal weighed to within 5 grams of target. The study's own authors acknowledge that long-term weight loss trials using nutrition education — where people choose their own food — find only a small advantage for low-carb diets.
The study lasted 20 weeks. Whether the metabolic effect would persist for a year or five years is unknown. A separate metabolic ward study found the energy boost started fading within four weeks.
The weight loss maintenance context matters too. These participants had recently lost significant weight, which changes metabolic rate. The same effect might not appear in someone who has been weight-stable for years and never dieted.
This is one study, not a settled consensus. It's the largest controlled-feeding trial on this question — 164 people for 20 weeks, with every meal prepared. That's a serious amount of data.
But two independent reanalyses challenged the core measurement. The debate over the actual size of the effect is active, published, and unresolved. Hall's critique uses the study's own data. Ebbeling's team published a formal defense.
The finding deserves attention, not certainty. It's strong enough to take seriously and honest enough to hold with open hands.
The carb-insulin model makes two big claims. This study tested the first: does staying low-carb make your body burn more energy? The answer was a fight, not a verdict.
The second claim — that cutting carbs makes you eat less without trying — got tested in a locked feeding ward at the NIH where twenty out of twenty people ate 689 extra daily calories on keto. The result didn't shrink under reanalysis. It flipped.
What This Study Found
All findings from this paper, in plain language.
- People who ate low-carb burned 209 extra calories per day compared to those who ate high-carb, with the same protein and total calories.
- Among participants who stuck closely to the diet, the low-carb group burned 278 extra calories per day — an even larger gap.
- People whose bodies produced the most insulin saw a 3.6 times bigger metabolic boost from low-carb than those who produced the least.
- Resting metabolism — the calories your body burns doing nothing — didn't change significantly between the diet groups.
- The hunger hormone ghrelin dropped more on the low-carb diet, which may help explain why some people feel less hungry eating fewer carbs.
- Leptin, a hormone that signals fullness, rose less on the low-carb diet among participants who stuck to the protocol — a possible sign of improved hormone sensitivity.
- The extra calorie burning didn't fade over the 20-week study — it was just as strong at week 20 as at week 10.
- The researchers projected that if the effect lasted, a typical man could lose an extra 10 kilograms over three years — but this was a mathematical estimate, not a tested outcome.
- Physical activity levels — how much people moved during the day — didn't differ significantly between the diet groups.
- Blood tests confirmed the diets were genuinely different — lipid markers and a carbohydrate-specific blood test all separated cleanly between groups.
- Body weight stayed stable during the test phase — less than 1 kilogram change on average — confirming the calorie differences happened during weight maintenance, not during active weight loss.
- After adjusting for age, sex, body composition, and starting metabolism, the low-carb group still burned 185 extra calories per day — smaller than the headline number but still significant.