What the research actually shows
Almost every diet study has the same fatal flaw: it relies on people telling you what they ate. They don't know, and they under-report by 30–50% without meaning to. So the honest way to answer this question is to lock people in a room where every gram of food is weighed and every breath is measured. Two studies did exactly that, and both were designed to give the low-carb answer a fair chance to win.
Same calories, same protein, same exercise. Cutting fat produced 89 g of body fat lost per day; cutting carbohydrate produced 53 g. Carbohydrate restriction did raise fat oxidation, exactly as the low-carb model predicts, and it still lost less fat, because burning fat and losing fat are not the same thing when you're also eating more of it.
The obvious objection is that six days is too short and the diet wasn't ketogenic. So the same lab ran it again, longer and in ketosis.
Four weeks of high-carbohydrate baseline, then four weeks of a ketogenic diet at identical calories and protein. Energy expenditure did rise on keto, by 57 to 151 kcal/day depending on the measurement method. That is a real metabolic advantage, and it is also the entire size of the effect. Meanwhile body fat loss slowed down during the ketogenic phase.
So the strongest version of the "calories don't matter" argument, tested under the strictest conditions available, produced an advantage worth about one banana a day, in the opposite direction from the fat loss it was supposed to explain.
Adults with obesity, all male in the 2016 study, confined to a metabolic ward for weeks. That cuts both ways. The control is unmatched (nobody forgot to log the olive oil), but these are not lean trained lifters, the diets ran for days to weeks rather than months, and adherence was enforced by the walls. What generalises is the physiology, not the compliance.
Why "calories don't count" keeps selling
Because diets that work do something else on the way. Cut carbohydrate hard and you lose several kilos of water and glycogen in the first week, drop most processed food by accident, and often eat more protein, which is filling. The scale moves fast, and the mechanism you're handed is "insulin", not "you ate 600 fewer calories without noticing."
This matters practically, because if you believe the macro did the work, you'll respond to a stall by cutting carbs further instead of looking at portion sizes. The stall is almost never metabolic . See what actually happens to your metabolism in a deficit.
The surplus side of the same coin
It runs the other way too. You cannot build meaningful muscle tissue out of nothing, which is why bulking exists. But the amount of surplus required is far smaller than the internet suggests, and the excess above that becomes fat with the same indifference . See how big a surplus you actually need.
What to do
- Set the deficit first, then argue about macros. Roughly 20–25% below maintenance is a sane starting point.
- Pick the diet you can adhere to. Low-carb, high-carb, and everything between produce the same fat loss at matched calories and protein. Adherence is the variable you actually control.
- Protect muscle with protein and lifting. Around 1.6–2.2 g/kg/day . See the protein page.
- If the scale stalls for 3+ weeks, audit intake before you audit your metabolism. Weigh a week's food honestly, including oils, drinks, and weekends.
- Cap the rate. Losing faster than about 1% of bodyweight a week costs you muscle . See how fast to lose.
Want this set up around your actual life?
I coach in person in St. Louis and online. First consultation is free.
Get in touch