The minimum effective dose of exercise for healthspan

Of the seven factors in this calculator, exercise is the only one with a substantial positive range. Not smoking doesn't earn you bonus years — it just avoids losing them. Exercise genuinely adds. That makes it the highest-value factor for most people, and it also makes the shape of its dose-response curve the single most useful thing to understand about it.

The shape of the dose-response curve

The relationship between physical activity and mortality is curvilinear with steeply diminishing returns. Plotted with activity on the x-axis and risk on the y, the line drops sharply at the left edge and then flattens out.

The clearest evidence comes from studies using accelerometers rather than questionnaires — a genuinely important methodological distinction, because people are poor at estimating their own activity and tend to overestimate it substantially. A 2019 meta-analysis by Ekelund and colleagues in the BMJ pooled accelerometer-measured data across multiple cohorts and found that risk reduction begins well below the commonly cited 150-minutes-per-week guideline, with the steepest part of the curve occurring at the very low end of activity.

This matters because the standard public health message — 150 minutes of moderate activity per week — is often heard as a threshold. It isn't. It's a reasonable target, but the data show meaningful benefit accruing long before you reach it, and the marginal benefit of the 140th minute is much smaller than that of the 20th.

Going from nothing to something

The practical implication is the most encouraging finding in this literature: the largest single gain available to a sedentary person comes from becoming slightly less sedentary.

Studies examining the least active groups consistently find that even modest amounts of light-to-moderate activity — a twenty-minute daily walk, or one or two short sessions a week — are associated with substantially lower mortality risk than near-total inactivity. The gap between "nothing" and "a little" is wider than the gap between "a moderate amount" and "a lot."

There's a reasonable mechanistic story for why. Complete physical inactivity produces a distinct constellation of effects: rapid loss of cardiorespiratory fitness, insulin resistance, muscle atrophy, reduced bone loading, and impaired vascular function. Small amounts of regular activity interrupt most of those processes. Going from moderate to high volume is refining an already-functional system; going from zero to moderate is turning it on.

Where the returns flatten

At the other end: the curve flattens substantially somewhere in the region of 300 minutes of moderate activity per week, and beyond that additional volume shows little further reduction in all-cause mortality in most analyses.

Whether very high volumes become actively harmful is genuinely contested. Some studies of extreme endurance athletes have reported elevated rates of atrial fibrillation and coronary artery calcification; others find no excess mortality at any observed volume. The honest summary is that any harm at high volumes is small and uncertain, while the flattening of benefit is well established. Either way, the implication for almost everyone reading this is the same: there is no healthspan reason to train like an endurance athlete, and the model does not reward doing so.

One asymmetry worth noting is that intensity appears to matter somewhat independently of volume. Vigorous activity is associated with additional benefit beyond the equivalent duration of moderate activity in several analyses. So for someone already active, raising intensity is generally a more efficient lever than adding hours.

What moderate intensity actually means

The word "moderate" causes a great deal of confusion, and the confusion systematically inflates people's self-assessment. The usable definitions:

The talk test is crude but it correlates reasonably with measured heart-rate zones and it requires no equipment. Its main value is diagnostic: a large fraction of what people log as moderate activity is, by this test, light. If you can comfortably sing through your workout, the model's "light" category is the honest answer.

This isn't a small correction. Because the dose-response curve is steepest at the low end, misclassifying light activity as moderate is exactly the error that leads someone to believe they're capturing benefits they aren't.

Why strength training matters for healthspan

Most population research on exercise has measured aerobic activity, because that's what questionnaires and accelerometers capture. But for healthspan specifically — as distinct from lifespan — resistance training has a strong claim to being underrated.

The reason is sarcopenia: the age-related loss of muscle mass and strength, which typically begins in the fourth decade and accelerates thereafter. Its consequences are precisely the ones that define the end of healthspan rather than the end of life: reduced walking speed, difficulty rising from a chair, loss of balance, falls, fracture, hospitalisation, and the loss of independence that often follows. Grip strength and gait speed are among the better simple predictors of functional decline and mortality in older adults.

Resistance training is the intervention with the best evidence for preserving muscle mass and strength into later life, and it works at essentially any age — trials in adults in their seventies and eighties consistently show meaningful strength gains. Bone mineral density responds to loading too, which matters for fracture risk.

Our model folds strength work into the general activity input rather than scoring it separately, which is a simplification we're not entirely happy with. If you're choosing what to add to an existing routine and you're over forty, the case for adding one or two resistance sessions is stronger than the case for adding more cardio, and the model doesn't currently express that.

Sitting, steps, and incidental movement

Two further findings that complicate the simple picture.

Sedentary time has some independent effect. Prolonged sitting is associated with adverse outcomes partly independently of structured exercise — the "active couch potato" pattern, where someone exercises for an hour and sits for eleven. The association attenuates considerably after adjusting for total activity, so it's weaker than headlines suggest, but breaking up long sitting periods appears to be worth something on its own.

Steps predict well and don't need to reach 10,000. Step-count studies consistently find mortality benefit accruing from around 4,000 steps a day upward, with the curve flattening somewhere between 7,000 and 9,000 for most age groups. The 10,000 figure has no particular scientific basis — it originated as a marketing target for a Japanese pedometer in the 1960s.

Both of these point the same way: total daily movement matters, not just the hour you label as exercise. Walking to a shop, taking stairs, and standing up periodically all count toward the same underlying quantity.

How the model scores it

The calculator takes two inputs — days per week and typical intensity — and combines them:

Three features of this matrix are worth reading deliberately. The zero row is flat — at no days of exercise, intensity is meaningless, and the penalty is the model's second-largest single negative after heavy smoking. The top-right corner plateaus at +3 rather than continuing to climb, reflecting the flattening of returns. And the biggest single step anywhere in the table is from the "none" row to the "1–2" row, which is the front-loaded curve made concrete.

If your current answer is "none," this factor is where your leverage is, regardless of what the rest of your inputs look like — with the single exception of active smoking, which is larger.