Race-day weather has long been the accepted variable in marathon performance, discussed in the same breath as pacing and fuelling. A new analysis published in Sports Medicine - Open argues that a second environmental factor has been sitting in plain sight and going largely unmeasured. Researchers led by James Donaldson at the University of Leicester examined 2,756,553 net finishing times from six World Marathon Majors between 2010 and 2024 and found that ambient nitrogen dioxide on race morning was associated with materially slower finishing times across the whole field.
The scale of the dataset is what gives the finding its weight. Berlin, Boston, Chicago, London, New York City and Tokyo supplied 75 marathon-year events, with race-day nitrogen dioxide and fine particulate concentrations drawn from the Copernicus Atmosphere Monitoring Service reanalysis. Linear mixed-effects models, adjusted for heat index, wind speed and long-term trends, found that each additional microgram per cubic metre of nitrogen dioxide was associated with 1.43 minutes of lost time in men and 1.56 minutes in women. Fine particulate matter, the pollutant that usually dominates public discussion of air quality, showed no statistically significant effect.
The most striking result is not the average but the distribution of it. Recreational runners experienced nitrogen dioxide effects roughly six times larger than elite runners, 2.12 minutes per microgram against 0.34, and the gap held even when expressed as a percentage of mean finishing time at 0.67 per cent against 0.18. Younger runners aged 18 to 29 were more affected than those over 60, at 2.17 minutes against 1.32. Kolmogorov-Smirnov testing confirmed that higher nitrogen dioxide shifted the entire finishing-time distribution to the right rather than merely dragging out the back of the field.
Why the mass field should suffer more than the professionals is not settled by the data, though the authors note the obvious candidate: exposure time. A runner spending four and a half hours on a city-centre course inhales considerably more polluted air than one who finishes in two hours and two minutes, and does so during the late-morning window when traffic-derived nitrogen dioxide in an urban core is typically at its highest. That the effect is specific to nitrogen dioxide rather than particulates points towards a traffic signature rather than a general haze, which is the part of the finding with the clearest policy implications.
For race organisers, the practical questions are about scheduling and routing rather than cancellation. Start times, the density of the wave structure and whether a course loops through a congested arterial road are all controllable in a way that the weather is not. For runners, the takeaway is more modest and more useful: a disappointing time in a big-city marathon may owe something to the air as well as to the training block, and the slower the runner, the more of it there is to breathe. It is also a reminder that the environmental conditions that matter most to the mass field are not always the ones the elite race is discussed in.