Exercise

Being fit and strong goes with larger differences in death rates than any drug has shown, though the fitness figures are associations. Heart-and-lung training and strength training are separate levers that add up, and an hour of training does not fully cancel a day spent sitting still.

Do 3–4 hours a week of conversational-pace cardio (150 minutes is the floor), one or two hard interval sessions and two or three strength sessions; train balance more with every decade; and break up long sitting with real movement. Most of the evidence that this lengthens life comes from large observational cohorts, so it shows associations; randomised trials back the effects on fitness, muscle, bone density and falls, and the sections below say which is which.

Why training beats anything in a pill bottle

Training works through several systems at once, and the two main kinds add up. In a cohort of 479,856 US adults, meeting the aerobic guideline alone tracked with about 29% lower all-cause mortality, the muscle-strengthening guideline alone with about 11%, and both together with about 40%, more than either on its own.[1] Cardio-only protects the heart but does little for muscle or the spine; lifting-only builds strength but not VO₂ max.

The benefit doesn't come from burning calories:

  • Arteries. Repeated high blood flow during exercise makes the cells lining the arteries produce more nitric oxide, which keeps the lining healthy and makes plaque less likely to form; that may be part of why fitness protects.[2]
  • Mitochondria. Sustained moderate effort switches on the muscle's low-fuel sensor, AMPK, and through PGC-1α the building of new mitochondria.
  • Muscle as a signalling organ. Contracting muscle releases myokines that improve blood-sugar handling and lower inflammation. A bone-to-brain hormone signal has been proposed from mouse work, but it is a hypothesis, not a reason to train.

Together, training pushes on several hallmarks of aging at once: mitochondrial dysfunction, deregulated nutrient sensing, chronic inflammation and stem-cell exhaustion, alongside vascular function. No drug currently in human trials does that.

VO₂ max: the strongest predictor, best read as a receipt

Cardiorespiratory fitness, measured as VO₂ max (the most oxygen your body can use at full effort), is among the most consistent mortality predictors in medicine, but every study behind that is observational. Each extra MET of fitness (one metabolic equivalent, 3.5 mL of oxygen per kilogram per minute) goes with about 13% lower all-cause mortality and about 15% lower risk of a coronary or cardiovascular event.[3] An umbrella review of 199 cohorts and 20.9 million observations put it at 11–17% per MET and extended the link to a dozen conditions, from several cancers to dementia; its authors grade the certainty very low to moderate.[4]

There is no clear ceiling, with caveats. In 122,007 adults referred for treadmill testing at the Cleveland Clinic, the least fit had about five times the mortality of the elite-fit, and the elite died about 23% less than the merely high-fit, an extra margin that held only in adults aged 70 or older and in those with high blood pressure.[5] The American Heart Association wants fitness treated as a clinical vital sign.[6]

Whether the number itself is a lever is unproven. One genetic study, using inherited variants as a natural experiment, found that activity level and lean body mass raise VO₂ max but that VO₂ max itself showed no causal link to longevity.[7] It is one study with a weak genetic tool, so the reading it supports, that a high VO₂ max is the receipt of the work rather than the work itself, is plausible but thin. VO₂ max covers the mortality data and why the trials have not shown that raising it adds years.

Holding on to fitness matters. Among 93,060 US veterans tested about six years apart, gaining fitness went with lower mortality, and in people with heart disease losing more than 2 METs went with 37–74% higher mortality, even for the fittest quarter; only the fittest quarter without heart disease showed no clear rise. VO₂ max falls about 10% a decade, largely regardless of activity, and years of hard training can cut that loss by up to about half in young and middle-aged men. In the Dallas Bed Rest Study, three weeks of strict bed rest at age 20 hurt five men's work capacity more than the thirty years of aging that followed.

How to raise it. What drives the gain is minutes spent close to your maximum. The classic session is the Norwegian 4×4: four minutes at 90–95% of maximum heart rate, three minutes of easy recovery, four times.[8] Once or twice a week is the usual longevity dose. Most interval formats work; very short all-out sprint protocols lag the longer intervals somewhat. In sedentary adults, intervals beat steady moderate exercise only slightly, on weak evidence. The popular 80/20 "polarised" split (about 80% easy, 20% hard) is not established as better than a "pyramidal" week with more moderate work; pooled trials find no meaningful difference. Timing hard sessions by heart-rate variability probably gives a small edge at most, on weak trials. Training VO₂ max has the protocols and a week-by-week plan.

A watch's VO₂ max estimate is probably least accurate for the very fit and, on thinner evidence, the very unfit; using it as a trend over months may work but is barely tested. Wearable VO₂ max estimates covers how far to trust it.

Resistance training: muscle, strength and power

Lifting is the best-evidenced way to keep the muscle, strength and independence that age takes away. In observational data, pooled cohorts tie resistance training to about 21% lower all-cause mortality,[9] and the dose is small. Death rates are lowest at just 30–60 minutes of muscle-strengthening activity a week, and the benefit becomes unclear beyond about 130–140 minutes.[10] More is not better for longevity; it is for performance, a different goal.

Controlled trials firmly show gains in muscle and mood. In type 2 diabetes, regular lifting probably lowers the three-month blood-sugar average (HbA1c) by about 0.4 percentage points. Low muscle mass predicts worse survival after surgery, including cancer surgery.

Power probably fades faster than strength. Power, how fast you produce force, is what catches a stumble. In a snapshot of 100 healthy adults aged 65 to 89, each year of age went with about 1–2% less strength but about 3.5% less leg power, a gap significant in men only.[11] In a 10-year cohort of about 3,900 adults aged 46–75, those with the lowest relative power died at roughly six to seven times the rate of the highest. Strength predicted death too, but less sharply, and only power improved a risk model already holding the usual clinical predictors.[12] Both measures were upper-body, in one Brazilian clinical exercise-testing cohort: read the multiplier as a direction, and a reason to add power work, not to drop strength work.

How to train. Two to three sessions a week, covering six movement patterns (squat, hinge, horizontal and vertical push, horizontal and vertical pull) plus core, single-leg, grip and calf work. Stop most sets one to three reps short of failure: strength gains barely depend on going closer, and muscle growth benefits only modestly. Muscle is roughly 20–60% stronger lowering than lifting, and in older adults eccentric-focused training improved get-up-and-walk and sit-to-stand tests more than conventional training, probably with less strain on the heart. Machines give results comparable to free weights and are often easier for frail or nervous beginners to start on. Eat 1.2–1.6 g of protein per kg of body weight a day, spread across meals.

Cold plunging soon after lifting may slightly blunt muscle growth and probably costs a little strength (a penalty not seen with whole-body immersion). Plunge on non-lifting days or leave a generous gap; no study has tested how long a gap is enough.

Resistance training covers the trials, power work, safety and protein; Resistance training methods covers equipment, training with no gym, and the best exercises for each muscle.

Zone 2: the base, not the whole programme

Most weekly aerobic volume should sit at conversational pace. Zone 2 is the hardest effort at which blood lactate stays near its resting level: the first lactate threshold, better defined as roughly 0.5 mmol/L above your own resting baseline than as a fixed 2 mmol/L, which varies widely between people.[13] In practice you can hold full sentences but not sing. Drifting up into zone 3 is the most common mistake recreational trainees make.

Continuous endurance training raised whole-muscle mitochondrial content by an average of 23% across 353 studies and 5,973 participants; interval and sprint training raised it about 27%, statistically indistinguishable but in a fraction of the hours.[14] That is why easy cardio alone isn't enough. A 2025 review titled "Much Ado About Zone 2" pointed out that the very high easy volumes in popular prescriptions come from professional cyclists, and at the few hours a week most adults have, easy work alone may not deliver enough stimulus.[15] Pair it with one or two interval sessions.

Can you overdo it? Not for dying early, as far as the cohorts show. But decades of very high endurance volume carry replicated signals for atrial fibrillation, an irregular heart rhythm, and for calcium in the heart arteries. How much training the atrial-fibrillation risk takes is unresolved, and nothing suggests worrying at ordinary training volumes.

Zone 2 covers finding the pace, the weekly dose, and what easy training does inside the muscle.

Balance and falls: train it long before you need it

Falls are the leading cause of injury death in adults over 65, and more than a quarter of over-65s living at home report a fall each year.[16] About 22% of people who break a hip are dead within a year,[17] and roughly half of survivors never regain their previous mobility or independence.[18]

Three cheap field tests predict survival in observational cohorts:

  • Standing on one leg for 10 seconds. Among 1,702 adults aged 51–75, those who couldn't were about 84% more likely to die over seven years.[19]
  • The sitting-rising test. Sit down on the floor from standing and get back up with as little support from hands, forearms, knees or the side of the leg as possible, scored 0–10. In 4,282 adults aged 46–75 followed for 12 years, scores of 0–4 went with roughly four times the natural-cause mortality and six times the cardiovascular mortality of a perfect 10.[20] Skip it if a fall would be serious for you.
  • Grip strength. In about 140,000 adults across 17 countries, each 5 kg less grip tracked with about 16% higher all-cause mortality, a stronger predictor than systolic blood pressure.[21]

These sort populations far better than individuals, and nobody has shown that improving a score extends life: read a result as a prompt to train.

A Cochrane review of 108 trials in over 23,000 older adults at home found exercise cut the rate of falls by about 23% and the number of people who fall by about 15%. Balance-and-functional training has the most certain evidence, and programmes combining balance with progressive resistance the largest single estimate; resistance training on its own is rated uncertain.[22] But pooling the exercise trials, the US Preventive Services Task Force found no significant reduction in fall-related fractures or deaths.[23]

Fast power work belongs on top of balance training, not in place of it. How heavy is disputed: the 2026 American College of Sports Medicine (ACSM) position stand puts the useful load at 30–70% of a one-rep maximum,[24] while an international consensus for older adults sets about 60% for upper-body and 80% for lower-body lifts; about 60% sits inside both and is a sensible default. Mobility and balance covers the field tests and the falls evidence; Training mobility and balance explains when to use which, along with balance progressions, slip-and-trip and step training, and the best-tested programmes (Tai Chi, Otago).

Bone: heavy loading is the best-tested way to build it

Low bone density predicts a shorter life, not just a broken hip, though what the link means is unclear. In one US cohort, people with osteoporosis had about 40% higher all-cause mortality than people with normal bone, while milder bone loss (osteopenia) showed no clear excess overall; in adults over 50 it carried about 43% higher mortality.[25] No single cause of death explained the excess, and fracture deaths were never analysed. Losing bone and muscle together, osteosarcopenia, carries about 53% higher mortality than not having the combination.[26]

"Walk for your bones" has aged poorly. Pooled walking trials find nothing at the lumbar spine and only a small gain at the femoral neck, and competitive swimmers have bone density no better than sedentary people. Vibration plates show a low-certainty signal at the total femur and none at the femoral neck or spine: an add-on for people who cannot load, not a substitute.

In the Australian LIFTMOR trial, postmenopausal women with osteopenia or osteoporosis did supervised deadlifts, overhead presses and back squats above 85% of their one-rep maximum, plus jumping drop landings, twice a week for eight months. They gained about 2.9% bone density at the lumbar spine and 0.3% at the femoral neck, with no fragility fractures.[27] The follow-on trial found about two-thirds of that spine gain, 1.9%, and no advantage at the total hip.[28] Exercise now has the outcome that matters: pooled across eleven controlled trials, it cut major osteoporotic fractures by about a quarter, though programmes that grew more intense did not do better.[29] Supervised, gradually built-up heavy loading was well tolerated in these screened women, even with fragile bone, though the trials were too small to count fractures.

The ACSM position stand recommends training all major muscle groups at least twice weekly, at around 80% of one-rep maximum for strength;[30] for bone, pooled trials find heavy loading helps at the hip but not the spine, though head-to-head with lighter loads the difference was not significant. An International Osteoporosis Foundation working group argues that long sedentary time harms bone regardless of training, while calling that evidence emerging;[31] a large Canadian cohort that adjusted for activity found no such effect.[32]

Bone density covers dosing impact and lifting, who should be careful, calcium, vitamin D and supplements, and the Trabecular Bone Score, which reads bone structure from a standard scan.

One workout doesn't cancel a day of sitting

Long sitting is linked to earlier death even in people who exercise, though the evidence is observational. In a 21-country cohort of more than 105,000 adults followed for a median 11 years, more than eight hours a day of sitting carried about 20% higher all-cause mortality and 21% more major cardiovascular events than less than four hours. Activity softened the link without removing it: among people who met the activity guidelines, eight or more hours of sitting still went with about 17–19% higher risk of death or a major cardiovascular event than in the group that sat least and was most active. The link was steeper in lower-income countries.[33] Pooling more than a million adults, it took about 60–75 minutes a day of moderate activity to largely cancel the extra risk of high sitting time (less so for heavy TV viewing), several times what a typical exerciser does.[34]

The biology points the same way, from rodents. Lipoprotein lipase, the muscle enzyme that pulls fat out of the bloodstream, holds steady for about four hours after a muscle stops contracting, is down by about half at six, and falls 90–95% with sustained inactivity, measured in rats and mice. In the brain, more daily sitting went with faster shrinkage of the hippocampus over seven years in one cohort, even after adjusting for exercise,[35] though a review of 33 studies calls the overall evidence limited and mixed.[36] A genetic study links more leisure screen time to faster-running epigenetic clocks, with skeletal muscle suggested as the tissue involved.

About 7,000 steps a day, not 10,000. The 10,000 figure was a 1960s pedometer marketing slogan. In a 2025 pooled analysis, 7,000 daily steps against 2,000 went with 47% lower all-cause mortality, 38% lower dementia, 37% lower cancer mortality and 14% lower type 2 diabetes, plus 47% lower cardiovascular mortality and 28% fewer falls on weaker evidence. The death-rate curve bends around 5,000–7,000.[37] There is no threshold below which steps don't count. Pooling about 40,000 adults in seven cohorts, if the least-active fifth added 5 minutes a day of moderate-to-vigorous activity, roughly 6% of deaths might be prevented, and if everyone except the most-active fifth did, about 10%; a separate analysis of a large British cohort found smaller effects.[38]

Move little and often. Get up roughly every half hour, and make one break an hour a genuine movement snack. In a 2026 randomised trial, 86 sedentary office workers who took a three-minute equipment-free exercise break every hour for 12 weeks had lower fasting and post-meal blood sugar, less insulin resistance, a slimmer waist and better self-rated productivity than colleagues who carried on as usual.[39] It was a single-site, unregistered trial with no walking-break comparison: encouraging, not settled. The 20-8-2 rule (twenty minutes sitting, eight standing, two moving) is a handy mnemonic for that frequency, but the ratio has never been tested. Moving against your desk posture in breaks (extending the upper back and hips, opening the chest) costs nothing, though no study has shown it changes a health outcome.

Night-time activity has also been linked to higher death rates, probably mostly through shift work and ill health: a reason to prefer daytime movement, not a proven harm.

Variety helps. In two US cohorts totalling 111,467 adults, doing a wider variety of activities tracked with about 19% lower all-cause mortality at the same total volume.[40]

Sitting covers the enzyme biology, brain and cancer links, standing desks and what to do at your desk.

Practical exercise prescription

A pragmatic week for a healthy midlife adult:

ModalityVolumeNotes
Zone 2 aerobic3–4 sessions × 45–60 min150 min a week is the floor; conversational pace: brisk walking, cycling, jogging.
VO₂ max intervals1–2 sessions × ~25 min4×4 minutes near maximum with 3 minutes easy between, or shorter intervals.
Resistance training2–3 sessions × 30–45 minCompound lifts (squat, hinge, press, pull). 2–4 sets × 3–6 reps for strength; 6–12 for size. Add fast power work. The mortality curve is captured in the first 30–60 min a week; the rest buys strength and muscle, which are their own reasons.
Mobility / balance10–15 min most daysBalance at least three days a week, made harder as it gets easy; hip mobility, single-leg work, thoracic rotation.
Break up sittingHourly through the workdayThree-minute movement snacks, about 7,000 daily steps.

Time budget: the structured week above comes to about 4–7.5 hours plus the daily mobility minutes, the same week Training VO₂ max lays out session by session; its minimum effective dose is about 3–5 hours. The remaining ~110 waking hours are the separate sitting lever.

Test yourself annually. The one-leg stance and the sitting-rising test take two minutes together; grip strength with an inexpensive hand dynamometer is the third.

What's overrated

  • The 10,000-step target. Most of the benefit arrives by about 7,000.
  • Walking alone for bone. Good for the heart, not for the spine; heavy lifting plus impact is the best-tested way to build bone.
  • Static stretching for injury prevention. It increases range of motion, but there's no good evidence it prevents injuries; a dynamic warm-up is the better default.
  • Ignoring recovery. Sleep, enough protein and managing chronic stress are part of training, not separate from it.

For the full ordered action list drawn from every pillar, see the healthspan long list.

— § —