What actually works in exercise

Pick up something heavy enough that the last repetition is genuinely difficult. Put it down. Do that again in a couple of days, with something slightly heavier.

That's most of it. The rest of this article is why it works, which bits of the surrounding noise hold up, and which bits have been repeated so often that people assume somebody must have checked.

Fitness advice arrives at a volume completely out of proportion to how much has actually changed. New methods appear every year. The underlying physiology hasn't moved in decades, and the honest summary of it is short, unglamorous, and considerably more useful than the marketing. None of what follows is medical advice, and anything to do with your own health, injuries or medication belongs with a doctor rather than an article.

Adaptation is the whole mechanism

Exercise doesn't build anything. That's the part people get backwards.

What a session does is impose a demand your body isn't currently equipped for. The building happens afterwards, over the following hours and days, as the body adjusts so that the same demand is less of an ordeal next time. Training is the stimulus. Recovery is where the result gets made.

Everything else falls out of that single idea. If the demand never rises, adaptation stops, because the body is already comfortably equal to it. If the demand rises faster than recovery can keep pace, you bank fatigue and niggling injuries instead of fitness. The whole craft of training is managing the gap between those two failures, and there's no clever way round it.

Which is why rest days aren't time off. They're when the training takes effect. Someone hammering seven days a week isn't training more than someone doing four — they're adapting less from the same work, and they're usually the one who ends up sitting out a month.

Progressive overload, which is dull and non-negotiable

The most reliable principle in the whole field is that the demand has to keep going up.

It doesn't have to be weight. More repetitions counts. More distance counts. Shorter rests between efforts, better control of the movement, a harder variation of the same exercise — all of them count. What doesn't count is repeating an identical session indefinitely and expecting the results to keep coming.

That's the standard reason people plateau after a promising start. The first couple of months feel spectacular because almost anything is a novel stress to an untrained body. Then the same routine stops being a stress at all, the body has no reason to change further, and the person concludes their genetics ran out.

The awkward corollary is that progress has to be measured. Not obsessively, and not with an app that grades your soul, but written down somewhere, because memory is unreliable and "it felt hard" is a mood rather than a measurement. People who keep any kind of record are startled by how often the number they were sure they'd been lifting is the number from four months ago.

You get what you train, and nothing adjacent

The second principle sports science agrees on is specificity, and it's less obvious than it sounds.

Adaptations are local and particular. They happen in the muscles that did the work, at roughly the speeds and ranges you worked through, in the energy systems you taxed. Train heavy and slow and you get better at heavy and slow. Train a long steady effort and you get better at long steady efforts. The transfer between them is real but modest, and it's much smaller than most people expect.

A very strong lifter can be badly out of breath on a flight of stairs. A marathon runner can find a heavy deadlift genuinely alarming. Neither is a paradox and neither is a moral failing — they've each adapted precisely to what they asked for.

The practical version: if there's something specific you want to be able to do, the training has to look like it at some point. General fitness carries you a long way, and then it stops.

Why the first fortnight feels like magic

Start lifting and you'll often be noticeably stronger within two weeks — much faster than muscle can plausibly be built.

That early jump is mostly nervous system. Your brain gets better at recruiting the fibres that already exist, firing them in a coordinated order, and switching off some of the protective inhibition that throttles an unfamiliar movement. You aren't growing much. You're learning to use what you've got.

Visible change in muscle size comes later and much more slowly, and that mismatch is behind an enormous amount of misplaced discouragement. Strength and size are related, but they're separate measurements moving on separate timescales — which is exactly why a smaller lifter can out-lift a bigger one, to the annoyance of the bigger one.

The same explanation covers something more cheerful. A long layoff costs less than people fear, because a good chunk of what you learned was learned by the nervous system, and it comes back far faster the second time. Detraining is real, but retraining is not starting from zero.

Strength and endurance adapt different systems

Resistance training and endurance work aren't two flavours of the same thing, and neither substitutes for the other.

Loading a muscle produces mechanical stress and small amounts of damage, and the repair response raises the muscle's capacity to produce force. It also loads the skeleton, and bone responds to load by getting denser — an idea usually traced to Julius Wolff in the eighteen-nineties and one of the few things that meaningfully affects bone density at all. That matters more the older you get, and it's the argument for resistance training that gets the least airtime.

Sustained aerobic work adapts a different set of machinery. The heart pushes more blood per beat, the vascular network improves, and muscles build more mitochondria and get better at pulling oxygen out of the blood and using it. Cardiorespiratory fitness turns out to be one of the strongest predictors of long-term health outcomes in large observational studies — stronger than several of the numbers people worry about far more.

For general health the answer isn't to pick one. It's to do some of each, which is what essentially every public health body recommends and what almost every argument on the internet manages to obscure.

The dose is smaller than the culture implies

The curve is steepest at the beginning. That's the single most encouraging finding in the whole area and the one least likely to be sold to you.

The difference between doing nothing and doing a modest amount is far larger than the difference between a moderate amount and a great deal. The returns keep coming as you add more, but they flatten. Most of the benefit is available at the bottom of the ladder.

Guidance from the World Health Organization lands at around a hundred and fifty minutes a week of moderate activity plus a couple of sessions involving muscle strengthening. That's a target, not a threshold. Being well under it is enormously better than being at zero, and a culture that presents training as an identity requiring several hours a week has quietly convinced a lot of people that half an hour is not worth starting.

What recovery actually consists of

Recovery is part of the programme rather than a gap in it, and most of it is boring.

Sleep is the largest single lever and there's no supplement that stands in for it. Past that, the useful principle is that hard and easy have to be balanced across the week rather than smoothed into every session. Hard days should be genuinely hard, easy days genuinely easy. What self-directed training gets wrong, more than anything else, is making everything moderate — too taxing to recover from properly, not taxing enough to drive much adaptation.

The soreness that shows up a day or two after unfamiliar exercise has its own name, delayed onset muscle soreness, and it isn't lactic acid. That's worth stating plainly because the myth is everywhere. Lactate clears from the blood within about an hour of stopping. Soreness peaks a day or two later, follows novel and particularly eccentric loading — the lowering half of a movement — and fades as you repeat the exercise.

Which means soreness is a poor guide to whether a session did anything. Training that's stopped making you sore is usually training that's working. Chasing the ache optimises for novelty rather than progress, and novelty is the enemy of the one principle that actually matters.

One honest caution: "listen to your body" covers both a real signal and a convenient excuse, and telling them apart is mostly a question of pattern. A run of missed sessions is information about your plan. A single reluctant Tuesday usually isn't.

Things that get repeated and don't hold up

Some claims survive purely on repetition. These are the ones worth naming.

• Spot reduction. Working a body part doesn't preferentially remove fat from it. Fat is mobilised from the whole body, and where it goes from first is largely down to individual physiology. Abdominal exercises build abdominal muscle. That's the entire list.
• Static stretching before exercise as injury prevention. The evidence for this is much weaker than its universality suggests, and long holds immediately beforehand can temporarily reduce force output. A warm-up that raises temperature and rehearses the movement you're about to do has better support behind it.
• Soreness as a measure of quality. Covered above. It tracks unfamiliarity, not effectiveness.
• Muscle turning into fat when you stop. They're different tissues and neither converts into the other. Muscle shrinks with disuse and fat can accumulate at the same time, which looks like a conversion and isn't one.
• Sweating out toxins. Sweat is a cooling system. How much you produce says more about the room than about the session.
• Lifting making people bulky by accident. Substantial muscle gain is slow and deliberate even for people actively chasing it. Nobody has ever woken up accidentally enormous.

One more that isn't a myth so much as a mismeasurement. Exercise on its own is a comparatively weak tool for weight loss. The energy cost of a session is smaller than most people estimate, and it's partly offset by moving around less for the rest of the day and by being hungrier. That isn't an argument against exercising — it's excellent for health, function and keeping the muscle you have. It's an argument against judging it by an outcome it was never particularly good at.

The supplement shelf is mostly weak

The supplement industry is enormous and the list of products with consistent support behind them is embarrassingly short.

Two things come up repeatedly in sports science with reasonably reliable results: caffeine, as a performance aid, and creatine, which is among the most studied compounds in the field. Almost everything else on the shelf rests on small trials, industry funding, surrogate measures, or nothing at all. That doesn't make any of it a recommendation — this is a description of where the literature is, not advice about what to put in your body, and that conversation belongs with a professional who knows your history.

Protein powder deserves a footnote, because it isn't really a supplement in the interesting sense. It's food in an expensive tub. What matters is total intake, and the powder is a convenience rather than an ingredient with special properties.

The general rule holds up well: the further a product's claims get from "this contains a thing you could have eaten", the thinner the evidence tends to be.

Two people, one programme, different results

Exercise trials keep finding something the marketing has no way to accommodate. Put a group on an identical programme for several months and the spread of outcomes is enormous. Some improve dramatically. Some barely shift. The average tells you very little about any individual in the group.

Nobody can predict in advance which you'll be, and the variation shows up separately for different qualities — a strong responder to endurance training isn't necessarily a strong responder to resistance training.

So when a programme "worked for" somebody, that's one data point drawn from a wide distribution, and it doesn't transfer to you with any reliability. The person selling it has no way of knowing whether they were the method or the outlier.

The research is shakier than the confidence around it

The field knows it has problems, and it's worth being blunt about them rather than pretending the citations settle things.

A large share of training studies run for eight or twelve weeks on a few dozen participants, very often young men, very often university students because that's who's available. Long-term data on how any of this plays out over decades is much rarer. Effects that look impressive over three months don't necessarily persist, and the outcomes that are easy to measure aren't always the ones that matter.

None of that makes the basics doubtful. Progressive overload and specificity are as close to settled as this field gets. It does mean that confident claims about the optimal number of sets, the ideal rest interval or the correct time of day to train are running well ahead of what anyone has actually demonstrated.

When you see a precise-sounding prescription, the honest question is how big the difference is between that and the obvious alternative. Usually the answer is: smaller than the argument about it.

The variable that decides it isn't in the programme

Almost every head-to-head comparison between two reasonable training approaches finds a smaller difference than the arguments would lead you to expect. Three sets against four. Full body against split. Free weights against machines. All of them work, and the gaps between them are modest.

Which means the variable that dominates isn't programme design. It's whether the thing happens at all, repeatedly, for years.

Adherence is a design problem, not a character trait, and treating it as a character trait is why so many people conclude they're the sort of person who doesn't stick at things. A slightly suboptimal routine you'll keep doing beats an optimal one you abandon in March, by a margin so wide it makes the original comparison irrelevant.

So convenience is a legitimate training variable. So is enjoyment. So is whether the plan survives a bad week, a late meeting or a fortnight of being ill without collapsing entirely. Choosing the gym you'll actually walk to, over the better one across town, isn't settling for less — on the evidence, it's the intervention with the largest effect available to you.

Everything else is detail, argued about loudly by people who'd get more from a walk.