Neither wins by much. In the largest year-long trial, adults on a healthy low-fat diet and adults on a healthy low-carb diet ended within 0.7 kg of each other on average, a gap small enough for chance to explain3. Pooled trials give low carb a small edge of about 1 kg, and ward studies show the reverse for body fat when calories are matched21.
In nutrition research, the practical difference between the two is smaller than the argument about them.
What was studied
The DIETFITS trial is the centrepiece3. It randomised 609 adults aged 18 to 50, with a BMI (body mass index, weight relative to height) between 28 and 40 and without diabetes, to one of two diets for 12 months. Randomising means a chance draw decided the diet, so the groups started alike.
Both groups attended 22 small-group sessions led by health educators. For the first 8 weeks, each group cut its target nutrient (fat or digestible carbohydrate) to 20 g a day, then added it back slowly to the lowest level each person felt they could keep for good3. Neither group had a calorie target. Both were told to eat plenty of vegetables and to minimise refined grains and added sugar. The authors called the diets "healthy low-fat" and "healthy low-carbohydrate" for that reason.
| What | Number |
|---|---|
| Adults | 609 (57% women), aged 18–50 |
| BMI range | 28 to 40, no diabetes |
| Length | 12 months, 22 group sessions |
| Completed | 481 (79%) |
| Calorie target | none |
Other evidence. A 2015 meta-analysis pooled 53 randomised trials of at least one year with 68,128 people, comparing low-fat diets with low-carb and other diets2. Two metabolic-ward studies, where every meal is measured, tested the two nutrients head to head in small groups14.
In plain terms: one big real-life trial, one big pooled analysis, and two tightly controlled lab studies.
What the studies found
What people actually ate. At 12 months, the low-fat group got 48% of calories from carbohydrate and 29% from fat; the low-carb group got 30% and 45%3. Protein was similar, 21% and 23%. Both groups reported eating about 500 to 600 kcal a day less than at the start, even without a calorie target.
Share of calories at 12 months, by diet group · %
Healthy low-fat Healthy low-carb
The result. Average weight change at 12 months was −5.3 kg in the low-fat group and −6.0 kg in the low-carb group3. The difference, 0.7 kg, had a 95% confidence interval of −0.2 to 1.6 kg. A confidence interval is the range the true value most likely sits in. Because it includes zero, the trial found no significant difference. The authors describe wide individual variability in weight change within the groups.
Genes and insulin did not sort people. The trial tested a pattern of three gene variants that had been proposed to predict who does better on which diet, and the amount of insulin released 30 minutes after a glucose drink3. Neither predicted who lost more on which diet.
The pooled picture. In weight-loss trials of a year or more, low-carb groups lost 1.15 kg more on average than low-fat groups (95% CI 0.52 to 1.79 kg)2. Against other higher-fat diets, low-fat diets showed no difference, and they did better only against people's usual diet. The authors concluded that against diets of similar intensity, the evidence does not support low-fat over other approaches.
Low-fat minus low-carb weight change, a year or more · kg
In plain terms: a year out, both roads end in nearly the same place, and low carb's small pooled edge is about 1 kg.
What the lab studies add
Ward studies can match calories exactly, which real-life trials cannot.
In a 2015 study, 19 adults with obesity spent two stays on a metabolic ward and received, in random order, a diet that cut carbohydrate and one that cut fat by the same number of calories1. Over 6 days, cutting carbohydrate raised fat oxidation (the rate the body uses fat as fuel) but led to about 53 g a day of body fat loss; cutting fat left fat oxidation unchanged and led to about 89 g a day. The authors' model predicted that over longer periods the body would narrow that difference.
In a 2021 ward study, 20 adults ate as much as they wanted of a plant-based low-fat diet and an animal-based ketogenic diet, 2 weeks each4. On the low-fat diet they ate about 689 kcal a day less over the two weeks. That is the opposite of what the idea that carbohydrate drives overeating predicts.
In plain terms: when calories are equal, cutting carbohydrate does not unlock extra fat loss; what differs in real life is how much people end up eating.
Why the arguments persist
Comments under diet videos say both "low carb >>>> low fat" and "consistency is the real key". The trials sit closer to the second. Both DIETFITS groups cut about the same number of calories, focused on vegetables and whole foods, and ended close together3. The authors suggest diet quality may explain why their result differs from older trials that compared low carb with less carefully built low-fat diets.
Short-term scale changes can also mislead: very low-carb diets shift stored glycogen and water early on (see why the scale jumps overnight). For how the macro split interacts with total calories, see macro splits compared and keto macros explained.
The studies at a glance
| Study | Who was studied | Key result | Year |
|---|---|---|---|
| Hall et al., *Cell Metab*1 | 19 adults with obesity, metabolic ward, 6 days per diet | Same calories: about 89 g/day body fat loss cutting fat vs 53 g/day cutting carbohydrate | 2015 |
| Tobias et al., *Lancet Diabetes Endocrinol*2 | 53 trials of 1 year or more, 68,128 people | Low carb 1.15 kg more loss than low fat; low fat no better than similar-intensity diets | 2015 |
| Gardner et al., *JAMA* (DIETFITS)3 | 609 adults with BMI 28–40, no diabetes, 12 months | −5.3 vs −6.0 kg group averages; 0.7 kg gap not significant; genes and insulin did not predict | 2018 |
| Hall et al., *Nat Med*4 | 20 adults, ward, eating freely, 2 weeks per diet | About 689 kcal/day less eaten on the low-fat diet | 2021 |
| Akbari et al., *Clin Nutr ESPEN*5 | 7 trials, 1,004 adults with overweight or obesity | Low carb ranked first for weight change; 2.7 kg more loss than Mediterranean; low carb vs low fat intervals wide and overlapping | 2024 |
Limits worth knowing
- Who was in DIETFITS. A well-educated area with good access to quality food; the authors flag this as a limit to generalising3.
- Self-reported intake. Food and activity in DIETFITS were self-reported; weight was measured on a clinic scale3.
- Dropout. 79% completed the year3.
- Ward studies are tiny and short. Nineteen and twenty people, days to weeks14.
- The pooled edge is heterogeneous. The 2015 meta-analysis mixed many diet designs across decades2.
- Abstract only. All studies except DIETFITS were read as abstracts; DIETFITS in full text, without its online appendix.
- People with diabetes were excluded from DIETFITS; for individual medical questions, a doctor is the right person to ask.
Latest research
The newest study we opened is a 2024 network meta-analysis of 7 randomised trials with 1,004 adults with overweight or obesity, comparing Mediterranean, low-carb and low-fat diets5. All diets led to weight loss. Low carb ranked first by probability, and in direct comparisons it lost 2.70 kg more than a Mediterranean diet (95% CI −4.65 to −0.75). Against a standard low-fat comparison diet, indirect estimates for low carb (−6.31 kg) and low fat (−5.61 kg) had wide, overlapping intervals. It is a small network, and a ranking is not a meaningful gap. It does not overturn the year-long head-to-head result.
The short version
- A year-long trial of 609 adults found healthy low-carb and low-fat diets about equal, 0.7 kg apart on average.
- Genes and insulin response did not predict who did better on which diet.
- Pooled long-term trials give low carb a small edge of about 1 kg.
- With calories matched on a ward, cutting fat produced as much or more body fat loss.
Educational overview only — not medical advice.
