Eating and drinking: what’s proven, what isn’t

Carbs during the race

More carbs help, up to a point, and then they stop helping.

One study tested twelve different doses, from 10 to 120 g per hour, four sites, double blind. Performance improved along a curve: about +3 % at 31 g per hour, +4 % at 48, +4.7 % at 78. Past that, very little ¹.

A review that looked only at running — which matters, because most of this research is on cyclists — found 13 of 17 studies showed a benefit, mostly in efforts over two hours ². The same review found gels made no difference over 16 to 21 km. Useful to know if you have been carrying one for a 10 km.

Duration decides everything here. Across 61 studies, the longer the effort, the bigger the gain ³. For an hour of running, what you eat barely matters. For four hours, it decides your race.

The days before

Carb loading fills the tank. Whether it makes you faster is less clear than you think.

The old three-day depletion phase turned out to be unnecessary. One day of eating a lot of carbs, with training stopped, does the job ⁷.

Then it gets uncomfortable. In a placebo-controlled trial, loading raised muscle glycogen as expected. But when riders also had breakfast and took carbs during the effort, the times were no different ⁸.

That does not mean skip it. It means loading and in-race fuelling overlap. If you fuel well during a long race, the week-long ritual beforehand is worth less than the internet suggests — and overdoing it has a real cost in your stomach.

How much to drink

Drink when you are thirsty. That is the answer the research supports.

The rule most runners absorbed — replace what you sweat, never drop more than 2 % of body weight — comes from lab protocols that look nothing like a race. Restrict the analysis to realistic conditions and the penalty largely disappears ⁹.

Body weight is a poor guide anyway. Over a long race you can finish several per cent lighter and still be fine ¹⁰.

When drinking to thirst was compared directly with drinking to a schedule, the schedule won nothing ¹¹. Thirst is not a compromise. It is the default the evidence supports.

Sweat rates vary enormously, from roughly 0.3 to over 5 litres an hour across people and conditions ¹². That is why no table can tell you your number ¹³.

Salt, and the one thing here that can kill you

Salt tablets do not protect you from this. In a randomised trial on 413 Ironman finishers, taking sodium made no significant difference to blood sodium compared with placebo or nothing ¹⁷.

Sodium is not pointless — sweat losses are real and vary a lot between people ¹². It is just a bad insurance policy against a problem caused by drinking too much. The protection is not more salt. It is not over-drinking.

Training your stomach

Practise eating at race rates, in training, before you need it.

Digestive trouble is the most common reason a good fuelling plan falls apart, and long efforts really do change how the gut works ¹⁸. The fix has some support: two weeks of practising at race intake reduced trouble and improved absorption in club runners ¹⁹²⁰.

One caveat you will not read elsewhere. The two studies most often cited as independent proof share authors and a nearly identical protocol. Treat them as one line of evidence, not two. And the progression every guide recommends — start at 60, build to 90 — has never actually been tested.

What is clear enough to act on: whatever you plan to eat and drink on race day should have been eaten and drunk at race effort, several times, first.

What changes in trail and ultra

The advice splits in two, and nobody has settled it.

The position paper written specifically for ultras says 30 to 50 g of carbs per hour ²¹. General endurance reviews say around 90 g for anything over three hours ⁶.

That is a factor of two to three, unresolved, in print. Anyone giving you one confident number for ultra fuelling has picked a side without telling you there is one.

Caffeine is the one supplement with a strong, consistent evidence base for endurance — with the usual caveats about how you personally respond, and about sleep ²².

Questions people actually ask