What this page establishes
- DIETFITS randomised 609 adults (BMI 28-40, no diabetes) to a healthy low-fat or healthy low-carbohydrate diet for 12 months with 22 group sessions each, and found no significant difference in 12-month weight loss, and no association with genotype pattern or insulin secretion.
- A secondary analysis of DIETFITS examined dietary adherence and diet quality as predictors of weight-loss success across both arms.
- In Hall's metabolic ward study, 17 men switched from a high-carbohydrate baseline to an isocaloric ketogenic diet with clamped protein; energy expenditure rose transiently by about 100 kcal/day in the first week, and the ketogenic diet was not accompanied by increased body fat loss.
- A dedicated review, 'Mystery or method? Evaluating claims of increased energy expenditure during a ketogenic diet', examines and largely does not support the metabolic-advantage claim.
Quick answers
If they all come out equal, does food quality not matter?
That is not what DIETFITS tested. Both arms were healthy versions, a healthy low-fat diet against a healthy low-carbohydrate diet, with 22 group sessions each.
Read the full answer
The trial says a well-constructed low-carb pattern and a well-constructed low-fat pattern land in the same place at 12 months. It says nothing about the junk-food version of either.I have type 2 diabetes. Does the no-winner finding apply to me?
DIETFITS enrolled adults with a BMI of 28 to 40 and no diabetes, so the headline null result does not settle the question for you.
Read the full answer
Within that population the trial did test baseline insulin secretion as a predictor and found it did not identify who would do better on which diet. Diet choice alongside glucose-lowering medication is a clinician conversation, because the interaction is with the medication rather than with the weight.These trials all came with group sessions. What do the numbers look like without that?
Lower, consistently. Each DIETFITS arm came with 22 group sessions, and PREDIMED-Plus paired its diet with a physical-activity programme, so the support is part of the intervention rather than packaging around it.
Read the full answer
Read every figure here as what is achievable under good conditions. What that support is actually delivering is covered at adherence and why diets fail.
The best diet question has an answer and it is boring. When frameworks are compared head to head with equal support, over enough time and with enough people, they land in the same place. That does not make the choice arbitrary, because the frameworks differ in who they suit and in what they do to your satiety and your lean mass. It just means you should choose based on your own failure pattern rather than on a claimed metabolic advantage, because the advantage keeps failing to appear when it is measured properly.
The trial that made this question boring: DIETFITS
Gardner and colleagues randomised 609 adults with a BMI of 28 to 40 and no diabetes to a healthy low-fat or a healthy low-carbohydrate diet for 12 months, with 22 group sessions in each arm. Published in JAMA in 2018.
There was no significant difference in 12-month weight loss between arms. The trial also tested two popular explanations for why individuals supposedly respond differently, a genotype pattern and baseline insulin secretion, and found neither predicted which diet worked better for a given person.
That last part is what makes DIETFITS the strongest evidence available on this question. It did not simply fail to find a winner. It went looking for the personalisation story that would explain the absence of a winner and did not find that either.
A secondary analysis examined dietary adherence and diet quality as predictors of weight-loss success across both arms, which is where the actual signal lives.
Low-carb and keto: what the metabolic-ward work found
The strongest claim for keto has always been a metabolic advantage: that the same calories produce more fat loss in ketosis. Hall tested it under the tightest available conditions, a metabolic ward where intake is provided and expenditure is measured rather than estimated.
Seventeen men switched from a high-carbohydrate baseline to an isocaloric ketogenic diet with protein clamped. Energy expenditure rose transiently by about 100 kcal a day in the first week, and the ketogenic diet was not accompanied by increased body fat loss. A dedicated review, "Mystery or method? Evaluating claims of increased energy expenditure during a ketogenic diet", examines the broader claim and largely does not support it.
So there is a small transient effect and it does not translate into more fat lost. Keto still works well for a lot of people, for a reason that has nothing to do with metabolism: cutting an entire food category reduces appetite and removes decisions.
Mediterranean: strongest for health, respectable for weight
PREDIMED-Plus tested an energy-restricted Mediterranean diet plus physical activity and found it effective at 12 months for reducing adiposity and improving cardiovascular risk factors in older adults with metabolic syndrome. Note the wording: energy-restricted. The classic Mediterranean pattern was studied for cardiovascular outcomes, not for weight, and it produces weight loss only when calories come down.
A body-composition analysis of the same trial in JAMA Network Open found the energy-reduced Mediterranean diet with physical activity significantly reduced total and visceral fat, and attenuated age-related lean-mass loss. That last part is unusual and it is the strongest argument for this framework over the others here.
High-protein: the one composition change with a mechanism that survives
High-protein eating is the exception to this page's general finding, and it is worth being precise about why. It does not beat other diets on total weight lost. It changes what is lost, by preserving lean mass under energy restriction, and it improves satiety per calorie, which helps adherence.
That is a real advantage and it is not a metabolic trick. It is also compatible with every other framework here: you can run a high-protein keto diet, a high-protein Mediterranean diet or a high-protein low-fat diet. Treat protein as a layer rather than a competing option.
Volumetrics and energy density
Volumetrics organises food by energy density rather than by macronutrient. High-water and high-fibre foods let you eat a normal-looking plate at a lower calorie total, so the strategy targets satiety directly.
It suits people whose problem is volume, the ones who feel unfed by a small portion regardless of what is in it. It suits people who eat by taste and by frequency much less well.
Low-fat: still works, no longer fashionable
Low-fat was the DIETFITS comparator and it performed identically to low-carb over twelve months. Its unpopularity is cultural rather than empirical.
It has a practical strength worth naming: fat is the most energy-dense macronutrient, so reducing it is a fast route to a lower calorie total without shrinking the volume of food on the plate. It works badly for people who find low-fat food unsatisfying, and that is a legitimate reason to pick something else.
| Framework | Weight-loss evidence | Best suited to |
|---|---|---|
| Low-carb / keto | Equal to low-fat at 12 months (DIETFITS), no metabolic advantage in ward conditions | People who want fewer decisions and get strong appetite suppression from it |
| Low-fat | Equal to low-carb at 12 months (DIETFITS) | People who eat by volume and do not miss fat |
| Mediterranean (energy-reduced) | Effective at 12 months in PREDIMED-Plus, plus visceral-fat and lean-mass benefits | People who want health outcomes alongside weight, and can cook |
| High-protein | Not more total loss, but preserves lean mass and improves satiety | Everyone, as a layer on top of another framework |
| Volumetrics | Works through energy density and satiety, no metabolic claim | People whose problem is portion volume |
How to choose based on your own failure pattern
Since the trials will not choose for you, choose on the mechanism of your last failure. If you quit because you were hungry on small portions, volumetrics or high-protein. If you quit because you were making too many decisions, keto or another rule-based framework that removes a category. If you quit because the food was miserable, Mediterranean. If you quit because you were tracking everything and hated it, a framework with rules instead of numbers.
Our /tools/stack-builder is useful at this stage and it is deliberately limited. It estimates a weekly rate of loss from the interventions you pick, and it will not let you add keto to 16:8 to calorie counting and sum the results, because those are three routes to one deficit. It takes the largest and shows you why on screen.
What all five have in common
Each one produces an energy deficit by a different route, and each has a population it fits and a population it does not. None has a metabolic advantage that survives measurement in a ward.
One more caveat applies to every number here. These effect sizes come from trials with group sessions, provided food or paid support. Free-living results are consistently lower, so read them as what is achievable under good conditions rather than as a forecast for your kitchen.
The evidence, one row per claim
| Claim | Tier | Source |
|---|---|---|
| DIETFITS randomised 609 adults (BMI 28-40, no diabetes) to a healthy low-fat or healthy low-carbohydrate diet for 12 months with 22 group sessions each, and found no significant difference in 12-month weight loss, and no association with genotype pattern or insulin secretion. | established | Gardner et al., JAMA 2018;319:667-679 |
| A secondary analysis of DIETFITS examined dietary adherence and diet quality as predictors of weight-loss success across both arms. | established | Am J Clin Nutr — DIETFITS adherence secondary analysis |
| In Hall's metabolic ward study, 17 men switched from a high-carbohydrate baseline to an isocaloric ketogenic diet with clamped protein; energy expenditure rose transiently by about 100 kcal/day in the first week, and the ketogenic diet was not accompanied by increased body fat loss. | established | Hall et al., Am J Clin Nutr 2016 |
| A dedicated review, 'Mystery or method? Evaluating claims of increased energy expenditure during a ketogenic diet', examines and largely does not support the metabolic-advantage claim. | established | PMC review |
| PREDIMED-Plus, a lifestyle intervention with an energy-restricted Mediterranean diet plus physical activity, was effective at 12 months in reducing adiposity and improving cardiovascular risk factors in older adults with metabolic syndrome. | established | Diabetes Care — PREDIMED-Plus one-year results |
| An energy-reduced Mediterranean diet with physical activity significantly reduced total and visceral fat and attenuated age-related lean-mass loss in adults with overweight/obesity and metabolic syndrome. | established | JAMA Network Open — PREDIMED-Plus body composition subgroup analysis |
| A secondary analysis of DIETFITS specifically compared ketogenic and ultra-low-fat dietary patterns on weight, insulin resistance, blood lipids and diet quality. | established | PMC — DIETFITS keto vs ultra-low-fat secondary analysis |
Sources
- Gardner et al., JAMA 2018;319:667-679
- Am J Clin Nutr — DIETFITS adherence secondary analysis
- Hall et al., Am J Clin Nutr 2016
- PMC review
- Diabetes Care — PREDIMED-Plus one-year results
- JAMA Network Open — PREDIMED-Plus body composition subgroup analysis
- PMC — DIETFITS keto vs ultra-low-fat secondary analysis
Keep reading
- Energy Balance: The One Mechanism Behind Every Method on This SiteThe mechanism page. Every other page here is a different way of arriving at the same physiology.
- Protein and Lifting: How to Lose Fat Without Losing MuscleThe most actionable page here. A deficit decides how much you lose. Protein and lifting decide what.
- Adherence: The Variable That Beats Every MethodThe strategy page. Design the plan around the way you fail, because the method itself is close to interchangea
- Losing Weight Without a GLP-1: What Actually Works, By the NumbersEvery non-drug method ranked by published effect size in one table, with the 'natural alternative to Ozempic'
- How Fast Should You Actually Lose Weight?Numeric and decision-shaped. A floor, a ceiling, and the reason both exist.