Carbs and fats: how much, and does the split matter?

The split between carbs and fat matters much less than most diet arguments suggest. In weight-loss trials, low-carb and balanced diets ended up about a kilogram apart after one to two years.^[1] Set protein first, keep fat at about 0.5 g per kg of body weight a day or more, and let carbs fill the rest.

Carbs (carbohydrates) and fat are two of the three macros, or macronutrients: the nutrients in food that supply calories. A calorie is a unit of food energy. Protein is the third macro, and it gets set first. The last article covered how much protein you need: about 1.6 g per kg of body weight a day for most people who lift.

Why isn't the answer obvious? Because both camps have a story about hormones and fat loss. The trials keep landing in the same boring place.

Why the split matters less than your total

What decides weight loss or gain is how much you eat compared with how much you burn (energy balance).^[2] Carbs and fat are two ways to fill that total. Protein and carbs carry about 4 calories per gram. Fat carries about 9.

Getting the split "wrong" rarely costs much fat loss, as the trials below show. Chasing the "right" one costs time and food you like. People drop bread, or cut every drop of oil, because someone online was sure. And in one trial, the people who stuck closer to their diet lost more, whatever the diet was.^[3]

There are two real limits. Fat can go too low. And carbs can matter for a long, hard session. Both are covered below.

What the research says

A meta-analysis pools the results of many studies on the same question into one estimate. In a randomised controlled trial (RCT), people are put into groups by chance, so the groups start out alike.

Low carb versus low fat: about the same weight loss

Moderate evidence. The review authors rated their own certainty as moderate, and most trials had a high risk of bias.^[1] That means flaws in how the trials were run could have skewed the results. A kilogram over a year or two is a small difference.

One large trial shows what that looks like. In DIETFITS, 609 adults ate either a healthy low-fat diet or a healthy low-carb diet for a year. Calories weren't set for either group.^[4]

The low-fat group lost 5.3 kg and the low-carb group lost 6.0 kg. The study couldn't tell those apart.^[4] That's one trial, but it matches the review above.

DIETFITS also tested two popular ideas: that your genes, or how much insulin your body releases after sugar, tell you which diet suits you. Insulin is a hormone that moves sugar from your blood into your cells. Neither predicted who did better on which diet.^[4] Limited evidence. One large trial.

A 2014 meta-analysis compared named diet programs across 48 RCTs and 7,286 adults. After a year, people on low-carb and on low-fat diets both lost about 7 kg more than people on no diet. The differences between named diets were small.^[5]

The authors wrote that this supports recommending any diet a person will stick to.^[5] Expert consensus. This is the authors' view, not a tested finding.

A smaller trial points the same way. In 2005, 160 adults were put on one of four popular diets for a year. People who said they stuck closer to their diet lost more weight, whatever the diet was. Which diet they were on didn't go with weight loss.^[3]

Limited evidence. Sticking to the diet was self-reported, and this is a link, not proof of cause. Only half to two-thirds of people finished.^[3]

When every calorie is controlled

A metabolic ward is a research unit where people live for days or weeks and every bite of food is measured. That lets researchers test whether the split itself changes fat loss.

In one ward study, 19 adults with obesity cut calories by 30% for six days, twice. Once the cut came from fat, once from carbs, with protein the same. Cutting fat removed about 89 g of body fat a day. Cutting carbs removed about 53 g.^[6]

Limited evidence. That's six days in a ward, and the gap is grams a day. The authors' model predicted the body would shrink that difference over longer periods.^[6]

A review that pooled 32 controlled feeding studies found the same direction. In those studies, carbs replaced fat at the same calories. People on the lower-fat diets burned about 26 more calories a day and lost about 16 g more fat a day.^[2] Moderate evidence for a small effect, measured over days to weeks.

Not every study agrees. One 2018 trial put 164 adults who had just lost weight on low-, medium- or high-carb diets for 20 weeks. The low-carb group burned about 209 more calories a day than the high-carb group.^[7] It measured calories burned, not fat lost.

Mixed evidence. One trial found a low-carb diet raised calorie burn after weight loss.^[7] Tightly controlled feeding studies that tracked fat balance didn't find more fat loss on low carb.^[2][6]

Carbs and your lifting

The worry runs the other way too: "no carbs, no gains". The research mostly says no.

A 2022 review of 49 studies looked at carbs and strength training.^[8] In single sessions, extra carbs helped in 6 studies and made no difference in 13. The benefit showed up mainly in two cases. Either the other group trained fasted (after many hours without food), or the workouts ran past 10 sets per muscle group.^[8]

Over weeks and months, 15 of 17 studies found strength results didn't depend on how many carbs people ate.^[8] Moderate evidence. Most of the studies were in people who already trained.^[8]

A 2022 meta-analysis of 21 studies tested carbs (mostly as drinks) around one session, against a placebo, a look-alike with no carbs (one study used plain water). With carbs, people got through more total work. The effect was clearest in sessions longer than 45 minutes and after a fast of 8 hours or more. In shorter sessions it wasn't clear.^[9]

Limited evidence. The authors rated the certainty as low, and 214 of the 226 people were men.^[9] It measured one session, not gains over time.

So carbs help a long, hard session, especially on an empty stomach. Your daily carb total mostly doesn't change your strength.

What about a daily number? For general fitness training, a review from the International Society of Sports Nutrition (ISSN) says a normal diet covers carb needs. That's about 3–5 g of carbs per kg.^[10] People training hard for 2–3 hours a day, 5–6 days a week, need more: around 5–8 g per kg.^[10] Expert consensus.

How low fat can go

Testosterone is the main male sex hormone. A 2021 meta-analysis pooled 6 small studies with 206 men. Low-fat diets (about 20% of calories from fat) lowered testosterone a little compared with higher-fat diets (about 40%).^[11]

The authors said more trials are needed to confirm it.^[11] Limited evidence. The men averaged 46 years old, and the studies lasted 2 to 10 weeks.^[11]

The review looked only at hormone levels, not muscle or strength.^[11] So it can't tell us whether that small drop changes your results in the gym. Nobody has studied this well yet.

The ISSN review suggests fat at about 30% of calories for athletes.^[10] When dieting, it says fat can go as low as 0.5–1 g per kg a day, which may be as little as 20% of calories.^[10] Expert consensus.

Carbs get a range, not a fixed number

Neither the European Food Safety Authority (EFSA) nor the World Health Organization (WHO) sets one carb number you must hit. EFSA says the exact need isn't known and depends on how much fat and protein you eat. It uses 45–60% of calories as a reference range.^[12] WHO gives no recommended carb amount, and says a range of intakes fits a healthy diet.^[13] Expert consensus. This is the panels' view, not a tested finding.

What WHO does recommend is where carbs come from: mainly whole grains, vegetables, fruit and pulses (beans, lentils and peas).^[13]

WHO also recommends adults get at least 25 g of fibre a day from food.^[13] Fibre is the part of plant food your gut can't fully digest. EFSA says 25 g a day is enough for normal bowel function in adults.^[12] Expert consensus.

The fibre number has data behind it. A 2019 analysis looked at 185 long-term studies of people and 58 trials.^[14] Compare the people who ate the most fibre with those who ate the least. The high-fibre eaters had 15–30% lower rates of early death, heart disease, stroke, type 2 diabetes and bowel cancer.^[14]

The drop in risk was largest at 25–29 g a day, and the authors' dose-response curves hinted that more could help further.^[14] Moderate evidence. Most of this comes from studies that watched people over time, so it shows a link, not proof that fibre caused it.

How to split your calories

The steps below follow the ranges above. The order and the default numbers go further than any trial. This is coaching practice, not a tested finding.

1. Know your daily calories. That's the energy balance step. An estimate is fine.
2. Set protein. About 1.6 g per kg of body weight a day. Lean and dieting: around 2 g per kg or more. Multiply by 4 to get its calories.
3. Set fat. About 0.5 g per kg at the least. A middle default is about 0.8 g per kg. Multiply by 9 to get its calories.
4. Give carbs the rest. Take protein and fat calories away from your total. Divide what's left by 4.
5. Pick carb foods with fibre. Whole grains, vegetables, fruit, beans and lentils. Aim for at least 25 g of fibre a day.
6. Eat some carbs before long or hard sessions. Especially if you train in the morning before eating, or your session runs past 45 minutes.
7. Then adjust to taste. Prefer more fat and fewer carbs, or the other way? Swap. Keep calories and protein the same.

Here's the sum for a 70 kg person (about 154 lb) eating 2,200 calories a day.

• Protein: 70 × 1.6 = 112 g. That's 448 calories.
• Fat: 70 × 0.8 = 56 g. That's 504 calories, about 23% of the total.
• Carbs: 2,200 − 448 − 504 = 1,248 calories. Divide by 4 and you get 312 g, about 4.5 g per kg.

Same person, liking fat more:

• Fat: 70 × 1.2 = 84 g, or 756 calories.
• Carbs: 2,200 − 448 − 756 = 996 calories, or 249 g.

Same calories, same protein. Based on the trials above, fat loss would be about the same either way.

Here's the fat floor at two other sizes. At 0.5 g per kg, a 55 kg person needs about 28 g of fat a day. A 110 kg person needs about 55 g.

At 0.5 g per kg, fat often lands below 20% of calories. That's at or below the level that lowered testosterone a little in the men studied.^[11] If that matters to you, stay nearer the 0.8 g per kg default.

Treat the numbers above as a starting point. The "best" split is the one you'll still be eating in three months. If you love pasta, a low-carb plan is a bad bet, whatever it promises.

If counting food makes you anxious, or you have a history of an eating disorder, skip the numbers and speak to a doctor or a registered dietitian.

You don't need to hit carbs and fat to the gram. Calories and protein do most of the work. Tracking food without obsessing covers how close is close enough.

Common mistakes

• Picking a side. Low-carb and low-fat fans both have studies to wave around. The biggest review found about a kilogram between low-carb and balanced diets, and a large trial found low-carb and low-fat about the same.^[1][4] Choose the foods you like and move on.
• Cutting fat to almost nothing. It saves calories fast, so it's tempting. Keep at least 0.5 g per kg a day. For a 70 kg person, that's 35 g.
• Fearing carbs for muscle. People hear that carbs spike insulin and store fat. In controlled studies, cutting carbs didn't add fat loss, and daily carbs mostly didn't change strength.^[2][6][8] Eat them if you like them.
• Matching your diet to your "type". In one large trial, a gene test and insulin response didn't predict who did better on which diet.^[4] Be wary of tests that claim to pick your diet from your genes.
• Counting carbs but not fibre. A carb target says nothing about the source. Fill it with whole grains, beans, fruit and vegetables, and the fibre adds up.

How Gymbers fits

Once protein is set, carbs and fat are just two ways to fill the rest of the day. The Gymbers app logs calories, protein, carbs and fat against your daily targets. Its targets are estimates, based on the Mifflin-St Jeor equation and an activity factor. Find a food by searching the food database or scanning a barcode, and add your own foods too.