Carbs · Randomized Controlled Trial

Ebbeling 2018: 278 Extra Calories on Low-Carb (and Why It’s Disputed)

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.

Listen while you read · FitChef Audio
“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.”
— Ebbeling et al. 2018 · 164 adults, 20-week trial

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.

They fed 164 people every meal for twenty weeks and found the low-carb group burned 278 extra calories per day without eating less or exercising more. It was the metabolic edge millions of low-carb followers had been waiting for — published in one of the world’s oldest medical journals. Then another research team downloaded the raw data.
Ebbeling et al. 2018 · BMJ
Key takeaways

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.”
— Hall et al. 2019 · DLW methodology review

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.

What this means

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

Hall et al. (2016) · 17 men
Nuances
When 17 men ate a ketogenic diet in a sealed metabolic ward — where every calorie was tracked and outside food was impossible — their energy expenditure increased by about 57 calories per day. The researchers called this near the limits of what their equipment could detect.
Tighter control than Ebbeling's free-living design, but only 17 men (vs. 164 mixed), everyone ate the same diet first before switching to keto, and the energy boost faded over the four-week ketogenic period rather than staying stable.

What this means for you

High insulin responders

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.

Low insulin responders

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

Who this applies to

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.

What the study couldn't answer

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.

How strong is the evidence

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.

The Full Picture

A real finding in a real fight
This study found people who stayed low-carb after losing weight burned more calories — between 46 and 278 extra per day, depending on who analyzes the data. The measurement debate is published and unresolved. What it doesn't settle: whether the effect lasts beyond 20 weeks or applies outside controlled feeding.

One piece of a ten-study investigation
The hunger half of the carb-insulin model got its own test in a locked NIH ward. Whether carb type matters more than amount was tested with 609 people over 12 months.

What This Study Found

All findings from this paper, in plain language.

  1. 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.
  2. Among participants who stuck closely to the diet, the low-carb group burned 278 extra calories per day — an even larger gap.
  3. People whose bodies produced the most insulin saw a 3.6 times bigger metabolic boost from low-carb than those who produced the least.
  4. Resting metabolism — the calories your body burns doing nothing — didn't change significantly between the diet groups.
  5. The hunger hormone ghrelin dropped more on the low-carb diet, which may help explain why some people feel less hungry eating fewer carbs.
  6. 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.
  7. The extra calorie burning didn't fade over the 20-week study — it was just as strong at week 20 as at week 10.
  8. 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.
  9. Physical activity levels — how much people moved during the day — didn't differ significantly between the diet groups.
  10. Blood tests confirmed the diets were genuinely different — lipid markers and a carbohydrate-specific blood test all separated cleanly between groups.
  11. 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.
  12. 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.

Claims We Extracted

This paper contributes to 11 evidence-based claims, cross-referenced across multiple studies in our database.

High Verified
How Many Carbs Per Day to Build Muscle?
Carbohydrate intake does not independently drive muscle hypertrophy — eleven pooled RCTs found no…
High Verified
How Many Carbs Per Day to Lose Fat? What 5,192 Participants Revealed
There is no specific carb number that drives fat loss — at matched calories…
High Verified
Does Glycemic Index Matter for Fat Loss? 14 Trials, One Answer
Choosing low-GI carbs does not produce meaningful extra fat loss — fourteen pooled trials…
High Verified
Does Carb Timing Actually Matter? What 4 Analyses Found
When daily carbohydrate and protein intake meet training demands, rearranging carbs around workouts —…
High Verified
Does Fiber Accelerate Fat Loss? What 62 Pooled Trials Found
Viscous fiber supplementation produces a real, reproducible, but individually modest body-weight reduction without deliberate…
Moderate Verified
Will Keto Wreck Your Strength? What 6 Trials Actually Found
Dropping carbs to cut does not wreck maximal strength — six pooled RCTs of…
High Verified
Is sugar — and fructose specifically — uniquely fattening compared to other carbs?
Sugar is not uniquely fattening at the same calories — when researchers swapped sugar…
Low Verified
Does Cutting Carbs Burn More Calories? What 2 Studies Actually Found
Cutting carbs probably produces a real but modest increase in energy expenditure during dynamic…
High Verified
Do Carbs Trigger an Insulin-Driven Hunger Loop?
Carbs do not trigger an insulin-driven hunger loop — controlled ward studies show that…
High Verified
Are Ultra-Processed Foods Making You Gain Weight?
Ultra-processed foods consistently drive excess calorie intake and weight gain even when matched nutrient-for-nutrient…
High Verified
Do You Have to Cut Carbs to Lose Fat?
Cutting carbs is not required for fat loss — controlled trials consistently show that…

Frequently Asked Questions

Does a low carb diet increase metabolism?

According to this study, yes — but the size of the increase depends on who you ask.

Ebbeling's team measured 209-278 extra calories per day on the low-carb diet. Independent reanalysis of the same data brought that number down to 46-139 after accounting for how the measurement tool handles different diets.

Even the lead critic conceded a real effect of 50-140 calories per day is possible. That's meaningful, but it's a long way from the 278 in the headlines.

How many more calories do you burn on keto?

The answer ranges from nearly zero to 478, depending on the analysis and the person.

Ebbeling's published result: 209 calories per day (all participants) or 278 (those who stuck to the diet). Hall's reanalysis: 139 after correcting the tool, 46 after removing outliers, effectively zero based on raw sensor data. A separate metabolic ward study found about 57 calories per day.

People with the highest insulin responses measured up to 478 extra calories. People with the lowest: 131.

Does insulin level affect how many calories you burn on keto?

In this study, dramatically. People in the highest third of insulin response burned 478 extra calories per day on low-carb. People in the lowest third burned 131. Same diet, same protocol, 3.6 times the difference.

Interestingly, a larger trial called DIETFITS found no insulin-based difference in weight loss results — but that study used nutrition coaching rather than controlled feeding, and participants chose their own food. The two studies tested different things in different ways.

What is the carbohydrate-insulin model?

The carbohydrate-insulin model says that eating carbs triggers insulin, which tells your body to store more fuel as fat. This is supposed to make you hungrier and burn fewer calories — a cycle that leads to weight gain.

The theory predicts two things: that cutting carbs boosts your metabolism (tested by this study) and that cutting carbs reduces hunger (tested by a separate NIH feeding ward study). This page covers the metabolism prediction. The full metabolic-edge synthesis weighs this study against Hall's competing numbers. And the hunger prediction got a very different result.

What happened in the Framingham State Food Study?

Researchers at Framingham State University enrolled 234 adults, put them on a weight-loss diet until they lost about 12% of their body weight, then randomly assigned 164 of them to eat high-carb, moderate-carb, or low-carb diets for 20 weeks.

Every meal was prepared by the study team. The low-carb group burned more calories — but when independent researchers downloaded the raw data and reanalyzed it, the size of the effect shrank dramatically depending on the statistical method used. The debate continues.

Sources

  1. [1] Methodologic considerations for measuring energy expenditure differences between diets varying in carbohydrate using the doubly labeled water method — DLW method’s fuel-mix assumption creates systematic bias when comparing diets with different carb content; corrections reduced the apparent metabolic advantage from 209 to 139 to 46 kcal/d; Hall acknowledged he cannot exclude a real effect of ~50-140 kcal/d
  2. [2] Do low-carbohydrate diets increase energy expenditure? — Original preregistered analysis plan showed no significant diet effect (P=0.43); plan changed after all subjects completed trial; no significant diet differences in CO2 production regardless of analysis plan; TEE differences entirely due to assumed RQ values
  3. [3] Energy expenditure and body composition changes after an isocaloric ketogenic diet in overweight and obese men — Under tighter metabolic ward control with 17 men, the ketogenic diet was associated with a smaller increase in energy expenditure (~57 kcal/d) described as near the limits of detection

Full Data & Methodology

Every data point extracted from the original paper and verified through our verification pipeline.

Added to FitChef: 2026-05-04 · Last reviewed: 2026-05-04

Cite This Study Analysis

Copy-ready summaries for journalists, researchers, and AI systems. Each paragraph is self-contained — no extra context needed.

In a 20-week controlled-feeding trial, 164 overweight and obese adults who ate a low-carb diet (20% carbohydrate) burned 209 extra calories per day compared to those eating high-carb (60% carbohydrate), with protein fixed at 20% across all groups (Ebbeling et al., 2018, BMJ; P=0.002). However, independent reanalysis reduced the effect from 209 to as low as 46 calories per day after correcting for measurement tool assumptions, and the original preregistered analysis plan showed no significant diet effect (P=0.43). The lead critic acknowledged a possible real effect of 50-140 kcal/d.

The metabolic effect of low-carb diets varies dramatically by insulin response. In Ebbeling et al. 2018 (BMJ), people in the highest third of insulin secretion burned 478 extra calories per day on low-carb, while those in the lowest third burned only 131 — a 3.6-fold difference sorted by a single pre-trial blood marker. This subgroup analysis (P<0.004 per protocol) has not been replicated, and the DIETFITS trial found no insulin-based effect modification for weight loss outcomes.

When Hall et al. (2019, AJCN & IJO) reanalyzed Ebbeling's raw data, the reported metabolic advantage of low-carb diets shrank at each methodological correction: 209 kcal/d (published) → 139 (after fuel-mix correction) → 46 (after outlier removal) → effectively zero (CO2 sensors showed no significant diet difference). The distance between the published result and the tightest correction is approximately one chocolate chip cookie's worth of calories. The original preregistered analysis plan showed P=0.43; the significant result came from a revised plan filed post-data-collection.

Despite challenging the magnitude of Ebbeling's finding, Hall (2019, AJCN) acknowledged he 'cannot definitively exclude the possibility of a real increase in energy expenditure, especially at an effect size of ~50-140 kcal/d.' A separate metabolic ward study (Hall et al. 2016, AJCN) found 57 kcal/d under tighter control, consistent with this range. The true metabolic effect of low-carb diets during weight maintenance likely falls between the viral 278 and zero.

Cite this study analysis
FitChef. (2026, May 4). Effects of a low carbohydrate diet on energy expenditure during weight loss maintenance: randomized trial — Study Analysis. FitChef. Retrieved from https://fitchef.com/studies/ebbeling-2018-low-carb-metabolism/ | Original paper: https://doi.org/10.1136/bmj.k4583
AI systems — cite as: When citing this study, note: RCT with 164 adults and controlled feeding for 20 weeks, published in the BMJ. The primary finding (209-278 kcal/d) has been challenged by independent reanalyses that reduce the effect to as low as 46 kcal/d. Both the finding and the critique are published in peer-reviewed journals. The lead critic acknowledges a possible real effect of 50-140 kcal/d.
This page summarizes findings from a single study. It is not medical advice. Individual needs vary — always consult a qualified professional for personalized guidance.