A study of 340 people who have each run an average of 121 marathons found aerobic capacity far above population norms — in the later age bands, up to 300% higher than the reference values. It is the kind of number that survives exactly one round of summarising before it becomes “marathon running keeps you 300% fitter as you age.”
The authors do not say that. They say close to the opposite, in a limitations section that is more useful than the finding.
The paper is VO2max ageing and all-cause mortality in a global cohort of multi-marathoners, by Leo Lundy, Richard Reilly and Neil Fleming at Trinity College Dublin, published in Scientific Reports in May 2026. It recruited adults who had completed — or were committed to completing — at least 100 marathons, across 24 countries, between January 2023 and March 2024. Final sample: 184 men and 156 women, mean age 52.2.
What it found
Across all eight age-by-sex subgroups between 30 and 69, the multi-marathoners exceeded the 50th percentile of the FRIEND reference database at p < .001. Against the 95th percentile — a much harder bar — three subgroups cleared it significantly: men aged 60–69, and women aged 40–49 and 50–69. Effect sizes ranged from Glass's Δ of 1.07 to 4.16, which is enormous.
The regression lines for the cohort also had higher intercepts and shallower negative slopes than the FRIEND curves, and the difference in slope was significant in every comparison. The population norm for VO2max decline is roughly 5–10% per decade after 30.
Why the authors won't let you draw the obvious conclusion
Four caveats do most of the damage, and all four are the authors' own.
The percentage is partly arithmetic. In their words, “modest absolute differences can yield disproportionately high percentage gains, which reflect mathematical artefact more than substantial physiological enhancement.” When the comparison group's VO2max is very low in the oldest bands, a real but moderate absolute advantage prints as a huge percentage. The 300% is at the top end and in the oldest groups — precisely where the denominator is smallest.
It is cross-sectional. Nobody was followed over time. A single snapshot of 70-year-olds who still run cannot distinguish slower decline from survivorship: the study has no data at all on people who took up multi-marathoning and stopped. The paper states it plainly — “these findings should not be interpreted as evidence of slowed physiological decline.”
VO2max was estimated, not measured. The values come from consumer wearables — Garmin, Polar, COROS — which the authors put at roughly ±10% against laboratory testing. They ran a sensitivity analysis at ±5%, ±10%, ±15% and ±20% and the findings held, which is the right thing to do. It is still not a metabolic cart.
The mortality figures are a model, not an observation. Nobody in this cohort died in the study. The mortality reductions come from applying a published linear conversion — about 3.7% lower all-cause mortality risk per 1 mL/kg/min of VO2max — to the measured differences. The authors call the results “directional group-level indicators, not precise individual-level risk estimates,” and warn that linear extrapolation “may exaggerate benefits at older ages.”
Add self-selected online recruitment, self-reported marathon counts, no baseline fitness or training history, and no control for socioeconomic status or underlying health, and the honest reading is narrow: people who have run 100+ marathons and are still running have exceptional aerobic capacity. Whether the running produced it, selected for it, or both, this design cannot say.
Our take: This is a well-run study that will be badly quoted, and the authors clearly knew it — the limitations section reads like a pre-emptive strike. The transferable skill is noticing that the headline number (300%) and the honest number (a real but ordinary-sized absolute advantage, direction unknown) come from the same dataset. When a fitness finding is expressed as a percentage above a reference group rather than an absolute value, check what the reference group's absolute value was. Small denominators manufacture spectacular percentages. That single habit will defuse most of what crosses your feed.
What to watch
- Longitudinal follow-up. The authors explicitly frame this as a framework for future research. Tracking the same cohort over years is the only design that answers the decline question.
- Measured VO2max. A lab-tested subsample would settle how much of the effect is real and how much is wearable estimation.
- Dropouts. The people who quit are the missing control group. Any study that finds them is worth more than this one.
- How it gets cited. Watch whether the 300% appears without its denominator. It usually will.
Worth saying: none of this makes the finding uninteresting. A 60-something running above the 95th percentile of a clinical reference population is remarkable, and the paper's framing of multi-marathoning as “a viable model for high-functioning ageing” is defensible. It just is not the same claim as “this slows your ageing,” and the authors are the ones drawing that line.