The basics of training, food and the brain are well researched; the claims made about them often go further than the evidence. This chapter keeps to the middle line between alarm and promise: what the guidelines say, what the large reviews find, how big the effects are and where the evidence ends. It covers how much movement is recommended and what the dose does, what exercise does for thinking, what creatine can and cannot do, and the basics of eating: protein, fibre, sugar, sweeteners and processed food. It informs; it does not replace advice from a doctor.
The middle line runs between two errors: taking every new study as proof, and taking every open question as a reason to do nothing. The figures in this chapter come from guidelines and from reviews that pool many studies. The ladder below sorts the main claims by how well they are supported. How large an effect is, is a separate question; each row names it.
Seventeen claims, sorted by how well they are supported. Tap a row for the figures and the source.
Well supported: consistent results from many studies, including randomised trials where they are possible. Probable: consistent results, mostly from observational studies or a few trials. Indications: few or small studies, or results that point in different directions. Open: contradictory or too little data. The classification is our summary of the reviews cited, using their own certainty ratings where they give one.
The guidelines are simple and have changed little in years. The WHO recommends that adults do 150 to 300 minutes of moderate or 75 to 150 minutes of vigorous aerobic activity a week, or a mix of both, and muscle-strengthening activity on two or more days. More brings additional benefits. Older adults should add training for balance and strength on three or more days. Long periods of sitting should be interrupted, although no threshold could be set. Austria's recommendations of 2020 use the same figures; Germany's of 2016 are being revised until 2027.
Move the sliders. Vigorous minutes count double, as in the WHO guideline. The step curve shows the data of 15 cohorts with 47,471 adults; it is an association, not a forecast for one person.
Relative risk of death during follow-up compared with the quarter of people who walked least (about 3,500 steps), Paluch et al. 2022. The points are the medians of the four groups; beyond the highest, 10,901 steps, the dashed line only carries the last value on. In a larger review of 57 studies (Ding et al. 2025) people walking 7,000 steps had a 47 per cent lower mortality than those walking 2,000, and the curve flattened from about 5,000 to 7,000 steps. For people over 60, Paluch found the benefit levelling off at 6,000 to 8,000 steps, for younger adults at 8,000 to 10,000.
Adults 18 and over, activity outside work, 2019, self-reported (European Health Interview Survey).
The principles on which the American College of Sports Medicine's position stand of 2026 is built, from 137 reviews with more than 30,000 participants.
The body adapts to what it is asked to do and then needs a little more: weight, repetitions, sets or speed. Without an increase, progress stops.
What is trained improves: heavy weights build strength, many sets build muscle, long efforts build endurance. Strength training does not replace endurance training, and the other way round.
Individualisation and adherence count as core principles: the best programme is the one a person keeps doing. Training to failure and elaborate periodisation did not consistently improve results.
One set is better than none, two are better than one. For strength: two to three sets at 80 per cent or more of the maximum, at least twice a week. For muscle growth: at least ten sets per muscle group a week; gains level off at around 18 to 20.
Trained men who did a single set two to three times a week gained on average 12.1 kg in their one-repetition maximum within 8 to 12 weeks: less than with more volume, but clearly measurable. Each extra weekly set adds less than the one before.
In 16 cohort studies, 30 to 60 minutes of muscle-strengthening activity a week went with a 10 to 20 per cent lower mortality. The curve is J-shaped: whether much more brings more is unclear.
Whether exercise makes people think better is a question with a history of overstatement. The effects exist, but they are smaller than headlines suggest, and in healthy adults they shrink when one corrects for the tendency of journals to publish positive results more often than negative ones. Two large reviews of recent years come to different conclusions; both are shown here.
Effect sizes in standard deviations with 95 per cent confidence intervals where published. 0.2 counts as small, 0.5 as medium, 0.8 as large. The switch shows two estimates after the authors' own corrections.
In a trial of 120 older adults, one year of aerobic training enlarged the hippocampus by about 2 per cent, while it shrank by about 1.4 per cent in the control group. Memory improved equally in both groups. Later reviews found no clear effect on the total volume.
In 58 cohort studies physically active people had a 20 per cent lower risk of dementia. Trials lasting up to five years found no effect on cognitive decline, and inactivity can also be an early sign of the disease. The Lancet Commission attributes about 2 per cent of dementia cases to physical inactivity.
A growth factor for nerve cells. A single bout raises its level in the blood (g 0.46), regular training raises the resting level a little (g 0.27). Whether this drives the effects on thinking is not proven.
Training increases blood flow in some brain regions, such as the hippocampus, but not in the brain as a whole, and it improves sleep quality moderately. New nerve cells in adult humans are disputed among researchers.
Interpretation, marked as such.
Read one way, exercise is the best-documented way without medication to support the brain: cheap, without side effects worth mentioning and with large benefits elsewhere. The review of 2,724 trials finds an effect across all ages and groups.
Read the other way, in healthy adults the measurable gain in thinking is small, and after correcting for unpublished null results it is close to zero. Most of the reviews pooled in the large estimate were rated critically low in quality. The great benefits of exercise lie in heart, metabolism, mortality and mood.
Both hold in part. Whoever trains for the body has good reasons; the effect on thinking is a possible bonus, not a promise.
Creatine is a small molecule that the body makes itself, 1 to 1.7 g a day, and takes in from meat and fish, 0.75 to 1.5 g a day in a mixed diet and almost none in a vegan one. About 95 per cent of it sits in the muscles, where it quickly supplies energy for short, intense efforts; a man of 70 kg stores about 120 g. As a supplement it is one of the most-studied products in sports nutrition. The evidence is strong for muscle and thin for the brain.
Effect sizes against placebo in standard deviations, with 95 per cent confidence intervals. Blue: muscle, orange: thinking, grey: mood.
In muscle the picture is consistent: with strength training creatine adds about 1.1 kg of lean mass and a small extra gain in strength, less in women and older adults. For memory the reviews find a small effect, almost entirely in older adults; in young people it is close to zero. The largest single trial (123 participants, 5 g a day) missed its main goals, and in 2024 the European Food Safety Authority found no proven cause-and-effect link with cognitive function.
"Creatine increases physical performance in successive bursts of short-term, high intensity exercise." Condition: 3 g a day, for adults doing high-intensity exercise.
"Daily creatine consumption can enhance the effect of resistance training on muscle strength in adults over the age of 55." Condition: 3 g a day and strength training at least three times a week over several weeks.
An application for a claim on cognitive function was assessed by EFSA in 2024: "a cause-and-effect relationship has not been established".
At usual doses no kidney damage was found in healthy people. The creatinine level in the blood rises, which can look like poor kidney function without being one; worth mentioning before a blood test.
With kidney disease or reduced kidney function, with medicines that strain the kidneys, with diabetes or high blood pressure, in pregnancy and while breastfeeding, for children and adolescents, and with bipolar disorder, where trials reported episodes of mania.
A loading phase typically adds 1 to 3 kg of body weight, mostly water. The fear of hair loss goes back to one study of 2009 with a rise in the hormone DHT; it was never replicated, and a trial of 2025 found no difference.
3 to 5 g a day fill the stores in three to four weeks; a loading phase of about 20 g a day for five to seven days does it faster, but the German Nutrition Society advises against high loading doses. After stopping, the stores return to normal within four to six weeks.
People with low stores, such as vegetarians, whose muscle stores are 20 to 30 per cent lower, gain more; some people with high stores hardly respond. In a study of 33 products, 44 per cent contained more than 100 mg per kg of creatinine, a breakdown product.
One study with 15 people found faster thinking after a night without sleep, with a single dose of 0.35 g per kg of body weight, five to eight times a usual daily dose. For depression the certainty of the evidence is very low.
On the basics, the bodies that write dietary guidelines agree: mostly plants, plenty of fibre, little free sugar, enough protein and little highly processed food. The German Nutrition Society's guidelines of 2024 recommend that more than three quarters of what we eat comes from plants. Austria's recommendations of 2024 list legumes as a group of their own for the first time, three times a week. The disputes are about details: how much protein, what sweeteners do, and whether processing itself harms.
The reference values for healthy adults, applied to your weight and energy intake. The value for strength training is the point at which, in a review of 49 trials, more protein brought hardly any more muscle.
Free sugar is all added sugar plus the sugar in honey, syrups and fruit juices; the sugar in whole fruit, vegetables and milk does not count. The European Food Safety Authority found that eating twice the protein reference value is safe for healthy adults; with kidney disease, protein needs medical advice.
0.8 g per kg of body weight under 65, 1.0 g from 65 (German, Austrian and Swiss reference values). With strength training, extra protein added 0.30 kg of fat-free mass in 49 trials, and the gain levelled off at 1.6 g per kg (confidence interval 1.0 to 2.2). For older adults an expert group recommends 1.0 to 1.2 g.
In 185 cohort studies and 58 trials, the risk reduction was largest at 25 to 29 g a day: mortality 15 per cent lower, coronary heart disease 24, type 2 diabetes and colorectal cancer 16 per cent. Austrian women eat 20 g, men 22 g; only 14 per cent reach the reference value.
The WHO recommends less than 10 per cent of energy as free sugar, ideally less than 5, about 25 g or six teaspoons. EFSA set no safe level: as low as possible. In Austria free sugar makes up 17.6 per cent of energy in women and 16.5 in men; more than four in five exceed 10 per cent.
Interpretation, marked as such: both sides stay.
Sweeteners should not be used to control weight or to prevent disease: they bring no lasting loss of body fat, and cohort studies link them to diabetes and heart disease, possibly distorted by other factors. A conditional recommendation, low certainty; it excludes people with diabetes.
Replacing sugary drinks with sweetened ones lowered weight by about 1 kg in 17 trials, and a trial of 2025 found slightly better weight maintenance after a year. For people with diabetes, diabetes societies accept sweeteners instead of sugar in moderation; water remains the first choice.
In 2023 the WHO's cancer agency classed aspartame as possibly carcinogenic, on limited evidence; that describes a hazard, not the risk at usual intake. The acceptable daily intake of 40 mg per kg stayed unchanged: a person of 70 kg would need more than 9 to 14 cans of diet soda a day. In September 2026 EFSA found no causal link to cancer.
A study of patients at high cardiac risk found more heart attacks and strokes among those with high blood levels (hazard ratio 1.8 to 2.2). The body also makes erythritol itself, so the measurement says little about diet; the US FDA saw no proof of cause. EFSA set an acceptable intake of 0.5 g per kg in 2023 and noted that, by its own probably high estimates, exposure can exceed it.
The NOVA system sorts foods by the degree and purpose of industrial processing. Ultra-processed products are industrial formulations with ingredients rarely used in kitchens, or with flavours, emulsifiers and sweeteners. In a controlled trial of 20 adults, people ate about 500 kcal more a day on an ultra-processed diet than on an unprocessed one, and gained 0.9 kg in two weeks instead of losing 0.9 kg. In a trial of 2025 with 55 adults, both diets met the UK guidelines; on the minimally processed diet weight fell by 2.1 per cent, on the ultra-processed one by 1.1.
Large reviews link a high share of ultra-processed food to cardiovascular death, type 2 diabetes and anxiety, but 41 of 45 analyses were rated low or very low in quality. The UK's scientific advisers point to confounding by other risks, critics to a classification that lumps very different foods together: in a European cohort, ultra-processed animal products and sweetened drinks were linked to more disease, ultra-processed bread and cereals were not. Whether processing itself harms, beyond calories and nutrients, is open. In Austria ultra-processed food makes up about 31 per cent of energy, in Germany about 38, in Europe on average 27.
The difference between two groups measured in standard deviations (SMD, Cohen's d, Hedges' g). By convention 0.2 counts as small, 0.5 as medium, 0.8 as large; an effect of 0.2 means that the two distributions overlap by about 92 per cent.
The range in which the true value lies with a stated certainty, usually 95 per cent. If it includes zero, the effect is not statistically certain.
Studies with positive results are published more often than studies without effect. Reviews that only pool published studies then overestimate effects; methods such as trim-and-fill try to correct for this.
A meta-analysis pools the results of many studies on one question; an umbrella review pools many meta-analyses. Their quality depends on the studies they include.
In a randomised trial chance decides who receives what, so the groups differ only in the intervention. A cohort study observes people over years; it shows associations, not causes.
The risk in one group divided by the risk in another. 0.80 means 20 per cent lower, 1.50 means 50 per cent higher. It says nothing about how large the risk is in absolute terms.
An international system for rating the certainty of evidence: high, moderate, low or very low. Observational studies start at low, randomised trials at high.
Metabolic equivalent: energy use as a multiple of rest, which is 1 MET. Moderate activity is defined as 3 to 6 METs, vigorous activity as more than 6. Fitness is often expressed in METs.
The heaviest weight that can be moved exactly once in an exercise (1RM). Training intensities are given as a percentage of it.
Creatine stores energy in the muscle as phosphocreatine. Creatinine is its breakdown product, measured in the blood to estimate kidney function; it rises when creatine is taken.
Acceptable daily intake: the amount of a substance per kg of body weight that can be taken every day for a lifetime without appreciable risk, set with a large safety margin.
Sugars added to food and drinks plus those naturally present in honey, syrups and fruit juices. Not included: the sugar in whole fruit, vegetables and milk.
Parts of plant foods that the small intestine cannot digest: in wholegrain, legumes, vegetables, fruit and nuts. They feed the gut bacteria, slow the rise of blood sugar and keep one full for longer.
The fourth of four groups in the NOVA system of the Brazilian researcher Carlos Monteiro: industrial formulations with ingredients rarely used in kitchens. The definition is criticised as too broad.