A new paper from the Garmin-RUNSAFE Running Health Study at Aarhus University in Denmark argues that the working model coaches and clinicians have used for decades to explain running-related overuse injuries is wrong on its central point: the injuries do not, in fact, accumulate gradually over weeks of cumulative load. They arrive suddenly, almost always inside a single training session, and the strongest predictor of whether that session is the one that breaks down is its length relative to the runner's longest run in the previous 30 days.
The study, which followed 5,205 runners across 87 countries for 18 months and logged 588,071 individual training sessions through Garmin device data, found that 35 per cent of participants sustained a running-related injury during the observation window. When the researchers reconstructed the timeline of each injury session by session, the conventional pattern of slow, accumulating tissue stress was almost never present. What they found instead was an exponential rise in injury risk on any single session whose distance exceeded 10 per cent of the longest run completed in the prior month — with the gradient of risk steepening sharply as the relative overshoot grew.
Lead author Rasmus Oestergaard Nielsen of Aarhus's Department of Public Health described the result as a paradigm shift, in part because the implication is that the most useful injury-prevention metric is not weekly volume, mileage trend, or any of the cumulative measures that running watches now report by default. The single most informative number, the paper argues, is the length of the runner's planned next session relative to their recent ceiling. Anything that significantly exceeds that ceiling on any individual day is the single best candidate for the day on which an overuse injury actually occurs.
The result re-frames, rather than overturns, the long-standing 10 per cent rule. The classic version of that rule applies to weekly mileage build-ups; the RUNSAFE finding applies to the distance of an individual session relative to that runner's recent longest. The two are not contradictory and, in many training plans, will produce similar advice. But for runners who carry a high-volume base with rare long efforts — the typical recreational marathoner who runs 40 miles a week with a 10-mile long run, then jumps to a 16-mile long run the weekend a plan starts — the session-level frame predicts an injury where the weekly frame does not.
The practical takeaways are unusually clean for an injury study of this size. Coaches building plans for masters-age and intermediate runners should now treat the 30-day longest-run figure as the binding constraint on any individual long session, rather than the weekly build, and should structure progression in long-run distance in 10 per cent increments week-on-week rather than allowing a one-step jump after a recovery block. For runners self-coaching, the same advice translates to a simple rule: never let a single run be more than 10 per cent longer than the longest run in the previous month, even when the calendar says you are due for a step-up.