How large is the difference in mortality risk between unfit and fit people?
Studies with tens of millions of observations show that fit people die roughly half as often as people with the poorest fitness, and that every improvement in endurance or muscle strength is linked to a measurably lower risk.
People with good fitness die far less often than people with poor fitness. In a large overview of more than 20.9 million observations, people with high fitness had a 53% lower chance of dying than people with the lowest fitness. An older but large analysis of nearly 103,000 participants confirms that picture: people with poor fitness (roughly the level of brisk walking) had a 70% higher chance of dying and a 56% higher chance of a heart or vascular event compared with the fitter group.
There is also a clear gradient: each additional step up in fitness is linked to an 11 to 17% lower chance of dying. That pattern has been found repeatedly across multiple large analyses. The strongest protective effect applies to heart failure: fit people develop it 69% less often.
Fitness also turns out to matter more than body weight. In research involving nearly 400,000 people, unfit individuals, regardless of whether they had a normal weight, overweight, or obesity, had roughly twice the risk of dying compared to fit people with a normal weight. Fit people who were overweight did not differ meaningfully from fit people with a normal weight. In other words, being fit with a few extra kilos is more favourable than being slim but unfit.
Muscle strength tells a similar story. People with a strong hand-grip had a 31% lower chance of dying than people with a weak grip, based on nearly 2 million participants. In women that difference was as high as 40%. Leg strength was also linked to a 14% lower chance of dying. These are all associations, not proven cause and effect: people who are fitter are also healthier in other ways, and that makes it difficult to isolate the effect of fitness itself precisely. But the size and consistency of the figures are impressive.
This answer draws on seven large combined analyses together covering tens of millions of observations. The researchers behind the most comprehensive overview rated the quality of the evidence themselves as low to moderate, mainly because all the underlying studies followed people over time without assigning anything; no one was randomly made fit or unfit. It therefore cannot be fully ruled out that other factors explain the difference. The direction and size of the effect are, however, consistent across virtually all large analyses.