How Much Carbohydrate Do You Actually Need?

The hard minimum: what the brain costs

The Institute of Medicine's 2005 reference intakes are the only place a real carbohydrate requirement is written down. They set an Estimated Average Requirement of 100 g a day and a Recommended Dietary Allowance of 130 g — the EAR plus two standard deviations at an assumed coefficient of variation of 15% — for children over one and for adults of every age. The basis is stated plainly: the average minimum amount of glucose used by the brain. At 4 kcal per gram, 130 g is 520 kcal, and that is the entire justification. Nothing about muscles, nothing about training, nothing about how you feel.

Even that floor is partly negotiable, and the same document says so. After a prolonged shortfall the brain adapts to run partly on ketone bodies — beta-hydroxybutyrate and acetoacetate — made in the liver from fat. That is why very-low-carbohydrate diets are survivable rather than lethal, and it is why the requirement is described as the amount of glucose the brain uses rather than the amount it cannot do without. The sparing is partial, not total: parts of the brain and the red blood cells have no alternative to glucose, which the liver will make from amino acids and glycerol if you do not eat it.

EFSA looked at the same literature and reached a different kind of conclusion. Its 2010 opinion sets no Average Requirement and no Population Reference Intake for carbohydrate at all, on the grounds that the absolute dietary requirement is not precisely known and depends on how much fat and protein are eaten alongside. It reports the 130 g figure, attributing it to the Institute of Medicine, and adds one number of its own: an intake of 50 to 100 g a day generally prevents ketosis. That difference in kind matters. One authority published a requirement; the other examined the evidence and declined to. Anyone quoting a single carbohydrate minimum as settled science is quoting only half the file.

The soft band: what the percentage ranges actually say

Above the minimum, both authorities stop talking about requirements and start talking about ranges. The Institute of Medicine's Acceptable Macronutrient Distribution Range for carbohydrate is 45 to 65% of energy; EFSA's reference intake range is 45 to 60%. These are not requirements and were never meant to be read as such. They are the spans within which a diet can meet its other obligations — enough protein, enough essential fat, enough micronutrients — without pushing any one macronutrient to an extreme associated with worse outcomes.

Turned into grams, the band is wide. At 1,800 kcal, 45 to 65% is 202 to 292 g. At 2,000 kcal it is 225 to 325 g. At 2,500 kcal it is 281 to 406 g, and at 3,000 kcal it is 338 to 488 g. The EU's narrower 45 to 60% ceiling shaves the top off each of those: 270 g at 1,800 kcal, 300 g at 2,000, 375 g at 2,500, 450 g at 3,000. Two things stand out. The band at any given intake is roughly 100 to 150 g wide, which is more than the entire published minimum. And the minimum sits far below the floor of the band: 130 g is 26% of a 2,000 kcal diet and only 17% of a 3,000 kcal one. EFSA makes the same point in its own arithmetic, putting 130 g at about 18% of energy for a man eating 2,800 kcal and 25% for a woman eating 2,100 kcal, and noting immediately that such intakes are not enough to meet energy needs once fat and protein are held to acceptable levels.

One unit note before the arithmetic goes further. Nutrition labels in the EU and the UK give energy in kilojoules as well as kilocalories, and some countries lead with the kilojoule. The conversion is fixed: 1 kcal is 4.184 kJ, so the 520 kcal in 130 g of carbohydrate is about 2,176 kJ. Regulation (EU) No 1169/2011 sets the labelling factor for carbohydrate at 4 kcal per gram, or 17 kJ. Every gram figure in this article can be moved between the two systems with that single multiplier.

If you train, the percentage frame breaks

The 2016 joint position stand of the Academy of Nutrition and Dietetics, Dietitians of Canada and the American College of Sports Medicine does not express carbohydrate as a percentage at all. It scales it to body mass and training hours, because that is what glycogen use scales to. Light days — low intensity or skill work — get 3 to 5 g per kilogram. A moderate programme of about an hour a day gets 5 to 7. An endurance programme of one to three hours a day at moderate-to-high intensity gets 6 to 10. Extreme commitment, more than four or five hours a day, gets 8 to 12. The table above turns those into grams for three body masses.

The two frames genuinely disagree, and it is worth seeing where. Take a 198 lb (90 kg) endurance athlete eating 3,500 kcal a day and training two hours. The training-load target is 6 to 10 g/kg, or 540 to 900 g. The percentage frame at 45% of 3,500 kcal gives 394 g — 4.4 g/kg, below the bottom of the band. Even the EU ceiling of 60% gives 525 g, or 5.8 g/kg, still short of the floor. A percentage that is correct for a sedentary person systematically underfeeds a heavy trainer, because a percentage of energy scales with appetite while glycogen depletion scales with hours and body mass.

The same position stand adds the timing figures that the daily total hides. During exercise, 30 to 60 g an hour covers muscle fuel and blood glucose; beyond two and a half to three hours, intakes up to 90 g an hour are associated with better performance, and only mixtures of glucose and fructose are absorbed fast enough to deliver that. A pre-event meal is 1 to 4 g per kilogram, one to four hours out. Refuelling between two demanding sessions less than eight hours apart runs at 1 to 1.2 g per kilogram per hour for the first four to six hours. None of those numbers can be recovered from a percentage of daily energy.

Glycogen: the tank that caps a single day

Carbohydrate you do not burn immediately goes into glycogen, and glycogen has a ceiling. A review in Nutrition Reviews puts whole-body glycogen at roughly 600 g, varying with body mass, diet, fitness and how recently you trained — about 2,400 kcal at 4 kcal per gram. Muscle holds most of it, at roughly 150 mmol per kilogram of wet weight after eight to twelve hours of rest, rising towards 200 in a fit, rested athlete after a few days of high-carbohydrate eating, and falling below 50 after prolonged intense exercise. The liver holds a separate store, averaging around 80 g in the same review's table, whose job is to keep replenishing the roughly 4 g of glucose circulating in the blood at any moment.

That ceiling is what caps the useful part of a single day's intake. Carbohydrate loading is written as 10 to 12 g per kilogram per day for 36 to 48 hours before an event over 90 minutes — for a 165 lb (75 kg) runner, 750 to 900 g a day. Notice that a single such day already exceeds the whole 600 g store. The protocol works because it runs over a day and a half while training tapers, so the store fills gradually and the surplus is burned or laid down slowly rather than banked in one sitting. Eating 900 g on one day, on top of a full tank, does not create a larger tank.

Which grams the label is counting

A target in grams is only as good as the definition of a gram of carbohydrate, and that definition is not the same on both sides of the Atlantic. A US label counts total carbohydrate by difference, with fibre inside the figure; an EU label counts metabolised carbohydrate only, with fibre on its own line. The same food therefore shows different carbohydrate numbers depending on where the pack was printed, and the popular net-carbs subtraction — carbohydrate minus fibre — is already done for you on one of them. Our companion piece on what the net-carbs subtraction assumes works through a full label; the arithmetic there is what turns any target on this page into shopping.

Fibre carries its own reference value, separate from the carbohydrate range. EFSA considers 25 g a day adequate for normal laxation in adults, and 2 g per megajoule for children over one year; it also notes evidence of benefit above 25 g, including lower risk of coronary heart disease and type 2 diabetes. Sugars get no upper limit at all in the same opinion — EFSA found the data insufficient to set one, while noting that frequent intake of sugar-containing foods raises caries risk. So the honest summary of the European position is a range for total carbohydrate, a floor for fibre, and a deliberate silence on added sugar.

What a number in grams cannot tell you

The calculator on this page will give you a range from your body mass and your training load, and that range is a real, sourced number. It is also silent on everything the sources are silent on. It does not know what those grams are made of — 300 g from oats, beans and fruit and 300 g from soft drinks meet the same target and do not behave the same. It does not know your glucose tolerance, and EFSA's opinion is explicit that the evidence on glycaemic index and glycaemic load was not strong enough to support a reference value. And it does not know whether you have a condition that changes the whole calculation.

That last point is the one that matters most. Diabetes, insulin therapy, epilepsy managed with a ketogenic diet, pregnancy, kidney disease and a long list of other situations move carbohydrate from a preference into a clinical variable. If any of them apply to you, the range on this page is background reading and nothing more; the number you act on has to come from the clinician who can see your own results.

Frequently asked questions

Is 130 g a day a minimum I have to hit?

It is a recommended allowance, not a hard floor, and its whole basis is the average minimum glucose the brain uses. The Institute of Medicine's underlying average requirement is 100 g; the 130 g adds two standard deviations to cover most people. Below that, the liver makes glucose from amino acids and glycerol and the brain shifts partly to ketone bodies, which is why very-low-carbohydrate diets are possible. EFSA declined to set a requirement at all and says only that 50 to 100 g a day generally prevents ketosis.

Why do the US and EU ranges differ, 45–65% against 45–60%?

Because they are the residue of two different judgements about fat and protein, not two different measurements of carbohydrate. Both ranges are built by fixing acceptable spans for the other macronutrients and letting carbohydrate take what remains, so a body that allows a slightly higher fat ceiling ends up with a slightly lower carbohydrate ceiling. The five-point gap is a policy difference, and at 2,000 kcal it is worth 25 g — smaller than the day-to-day variation in most people's intake.

I do not train. Which number applies to me?

The percentage range, converted at your own energy intake — and the honest reading is that it is a wide band rather than a target. At 2,000 kcal it runs from 225 g to 325 g under the US range, or to 300 g under the EU one. Where you sit inside that band is a matter of what you like to eat and what keeps you full, not of a physiological requirement, because the requirement was already satisfied at 130 g. The light training-load row of the table, 3 to 5 g per kilogram, is the closest thing to a sedentary figure the sports literature publishes.

Can I store extra carbohydrate for a race by eating a huge amount the night before?

Not usefully, because the tank has a size. Whole-body glycogen is roughly 600 g, about 2,400 kcal, and a supercompensated muscle holds around 200 mmol per kilogram of wet weight against a rested 150. The published loading protocol is 10 to 12 g per kilogram per day for 36 to 48 hours, not one enormous meal — it works because it spreads the load over a day and a half while training tapers, letting the store fill gradually. A single 900 g evening does not build a bigger tank; it mostly builds an uncomfortable night.

Does the type of carbohydrate change the target?

It changes what the target is worth, not the target itself. EFSA reviewed glycaemic index and glycaemic load and concluded the evidence was not sufficient to set a reference value, so no authority publishes a target in low-GI grams. What the same opinion does publish is a separate fibre figure — 25 g a day for adults — and a note that intakes above that are associated with lower risk of coronary heart disease and type 2 diabetes. In practice that is the lever: hit the carbohydrate range with foods that also carry the fibre, and the quality question largely answers itself.