Forget the 10% Weekly Rule: The 5,205-Runner Study

Almost every age-group triathlete has been taught the “10% rule” — don't increase your weekly running mileage by more than 10% — and, more recently, to keep an eye on an app's acute:chronic workload ratio (ACWR), the little traffic-light number that's supposed to flag when your training load is spiking dangerously. Both have been repeated so often they feel like settled science. A landmark 2025 cohort study of 5,205 runners suggests neither one is where the real risk signal lives — and points instead at something much simpler and much easier to check: how far today's run is compared with the longest run you've done in the last month.

This is a myth-busting story, but it needs careful handling. The study is genuinely large and well-designed, and its critique of ACWR lines up with a separate, well-established body of methodological criticism. But it is one observational cohort, it measured association rather than proof of cause, and it does not license a new blanket rule to replace the old one. Here is what it actually found, what it doesn't prove, and how to use it sensibly.

Context: the rules age-groupers already follow

The 10% rule is a simple heuristic: don't let your weekly running distance climb by more than roughly a tenth from one week to the next. ACWR is a more modern refinement, popularised through GPS watches and training-load apps, that compares your recent (“acute,” often 7-day) load against your longer-term (“chronic,” often 28-day rolling average) load, with ratios outside a supposedly safe band flagged as risky. Triathletes lean on both heuristics heavily because running injuries are the overwhelming majority of what sidelines them — overuse injuries concentrated in the lower limb (Journal of Sport Rehabilitation, 2025) — and because triathlon training almost always builds toward one weekend long run that jumps well beyond anything done earlier in the month.

Both heuristics are appealing because they're simple and because they promise control over something that genuinely worries age-group triathletes: getting hurt mid-build and losing months of race preparation to a stress fracture or tendon problem. That worry is legitimate — running injury rates in this population are high enough that a large, well-designed study challenging the standard advice is worth taking seriously, provided the new information is handled with the same care the old advice never quite got.

What the evidence actually says

The 5,205-runner cohort: what it measured and what it found

A 2025 study in the British Journal of Sports Medicine followed 5,205 runners (mean age 45.8 ± 10.4, 22% female) through 588,071 recorded sessions over up to 18 months, using Garmin-RUNSAFE device data rather than self-reported training logs (BJSM 59(17):1203–1210, 2025; open-access PMC version with full hazard ratios). The design is genuinely impressive for this kind of question: prospective, device-measured training exposure, 35% of runners reporting an injury over the study period (72% of those overuse), a cumulative overuse-injury probability of 30.5% (95% CI 29.0–32.0) by 200 sessions, and a multistate Cox model that treated traumatic injury as a competing risk rather than ignoring it (PubMed record with exposure-category definitions and hazard rate ratios).

The headline finding: neither week-to-week change in running distance nor ACWR was significantly associated with overuse injury in this cohort. What was associated, in a clear dose–response pattern, was the distance of a single session relative to the longest run completed in the previous 30 days. A run 10–30% longer than that reference point carried an adjusted hazard rate ratio of 1.64 (95% CI 1.31–2.05); 30–100% longer, 1.52 (1.16–2.00); and more than double the prior longest run, 2.28 (1.50–3.48) (BJSM, 2025). In other words, the risk climbed with the size of the single-session jump, and the pattern held even though the classic weekly-total metrics showed nothing.

Why this isn't a slam-dunk replacement rule

The study's own authors describe their findings as exploratory, and several features of the cohort matter: injuries were self-reported even though training exposure was device-measured, the sample was 78% male and drawn mostly from European and North American Garmin users, and the whole design is observational — it establishes an association and a dose-response gradient, not a proven causal mechanism. That means it does not license a precise “never exceed X% of your longest run” prescription, only a defensible, evidence-informed heuristic. Some media coverage has leaned into “dramatic” framing of the findings; that's journalistic emphasis, not a description of the actual effect sizes reported in the paper (Runner's World; Luxembourg Institute of Health release).

The case against ACWR specifically

The critique of ACWR as a metric is on firmer ground than the single-session finding, because it's backed by separate methodological work outside this one cohort. A widely cited 2020 paper laid out conceptual and statistical pitfalls in how ACWR is calculated and used, including the lack of proper causal estimation and reclassification artefacts introduced by the ratio itself (Impellizzeri et al., IJSPP 15(6):907–913, 2020). A 2021 reanalysis pushed further: dividing real acute training load by randomly generated, made-up “chronic” loads reproduced comparable odds ratios (1.16–2.07) to using the real chronic load — and neither the real ACWR nor acute load alone outperformed a model with no load information at all (c-statistic of 0.574, barely better than chance) (Sports Medicine, 2021). That's a serious problem for a metric millions of training apps display as a meaningful risk signal.

What we know — and don't — about triathletes specifically

Triathlon-specific injury data exists, but it's weaker in design than the 5,205-runner cohort. A survey of 219 injured triathletes found 58.5% of injuries were overuse and 73.8% affected the lower extremity, with 87% of athletes having to miss or modify training as a result (Journal of Sport Rehabilitation, 2025); a separate cross-sectional study of 758 Brazilian athletes examined time-loss musculoskeletal injuries with similar patterns (BMC Sports Science, Medicine and Rehabilitation 18:20, 2025). Both are survey-based or cross-sectional, which is a weaker design than the prospective, device-measured cohort above. And it's worth being explicit about a gap: the 5,205-runner study is running-only data. It cannot tell you how cycling volume interacts with a run-session spike, and triathletes' total training load includes swim and bike work the study never measured. That interaction is genuinely unknown, not just underexplored.

Practical application: replacing two rules with one heuristic

Stop tracking your weekly running-distance percentage increase as the primary safety check. This cohort found no significant association between week-to-week change and overuse injury — it isn't where the signal lives.

Stop treating your app's ACWR score as a validated red/amber/green indicator. The methodological critique here is strong: a metric that can't beat a model built on randomly generated numbers (Sports Medicine, 2021) is not a reliable basis for deciding whether to run today.

Instead, check a single number before a long or hard run: how does today's planned distance compare with the longest run you've completed in the past 30 days? Treat anything more than roughly 10% longer as worth a moment's thought, and anything more than 30% longer — or more than double — as a materially higher-risk jump, based on the dose-response pattern in this cohort.

Build and protect a genuine long-run habit rather than relying on occasional big efforts. A steady, recurring long run that creeps up gradually keeps your “longest run in 30 days” reference point moving with you, instead of leaving a large gap for any single session to jump across.

Remember this is a heuristic drawn from an association in observational data, not a guarantee. Use it alongside how your body actually feels — recovery, soreness, sleep, and mood — not as a replacement for paying attention to those signals.

Common mistakes

Fixating on a weekly percentage while ignoring one large long run buried inside that week. The week-to-week total told this cohort nothing; the single session did.

Trusting an ACWR traffic light as if it were a validated clinical tool. The reanalysis showing random “chronic load” numbers reproduce similar odds ratios to real training data (Sports Medicine, 2021) is a genuinely damning result for that specific metric.

Applying a running-only heuristic to total triathlon load without acknowledging the gap. Running on tired legs off a heavy bike block, or ramping up run volume right after a swim-and-bike-heavy training phase, are plausible compounding factors — but this study cannot tell you how large that effect is, so don't present it as measured.

Treating a single-session distance rule as a diagnosis rather than a probability. A run within the “safe” zone can still end in injury, and a run outside it does not guarantee one — this is a dose-response pattern in a population, not a personal verdict.

Ignoring pain that needs professional assessment because “the load numbers looked fine.” Persistent, worsening, or localised pain — especially bone pain, night pain, pain that worsens through a run, or an inability to run without limping — needs a physiotherapist, sports physician, or doctor, not a load-management app. Suspected bone stress injury requires prompt medical assessment regardless of what your training log shows.

Assuming this evidence applies equally to everyone. The cohort was 78% male and drawn mostly from European and North American Garmin users, and women are underrepresented in this literature generally. If you're returning from injury, illness, pregnancy or postpartum, or managing bone-health risk factors including low energy availability, use individualised professional guidance rather than a population-level heuristic.

How to apply this week

Look back at your training log and identify the longest single run you've completed in the past 30 days. That's your new reference point, not last week's total mileage.

Before your next long run, compare planned distance against that reference point. If it's within about 10% longer, that's the lowest-risk zone this cohort identified; if it's 30% or more beyond it, treat the jump as a deliberate decision, not an accident of the calendar, and consider spreading the increase across two or more weeks instead.

Turn off or stop acting on ACWR alerts from your training app this week, and don't let a green or red status change your plans on its own.

Keep a simple weekly note of how your longest run felt — effort, next-day soreness, sleep — next to the distance, so you're combining the load heuristic with how your body is actually responding.

If you're building toward a race and need to extend your long run meaningfully, do it in small, planned increments across several weeks rather than one jump, and treat any new localised or worsening pain as a stop signal, not a training-load calculation to solve.

References

• How much running is too much? Identifying high-risk running sessions in a 5200-person cohort study — British Journal of Sports Medicine 59(17):1203–1210, 2025
• Same study, open-access PMC version with full hazard ratios — BJSM, 2025
• PubMed record with exposure-category definitions and hazard rate ratios — BJSM, 2025
• Acute:chronic workload ratio: conceptual issues and fundamental pitfalls — International Journal of Sports Physiology and Performance 15(6):907–913, 2020 (Impellizzeri et al.)
• What role do chronic workloads play in the acute:chronic workload ratio? Time to dismiss ACWR and its underlying theory — Sports Medicine, 2021
• Injury epidemiology and preventative strategies in triathletes: a study of the 2022 season — Journal of Sport Rehabilitation, 2025
• Prevalence and predictive factors of musculoskeletal injuries in triathletes: a cross-sectional study of 758 athletes — BMC Sports Science, Medicine and Rehabilitation 18:20, 2025
• Rethinking the “golden” 10% mileage increase — Runner's World, 2026
• New research challenges decades of assumptions about running injuries — Luxembourg Institute of Health, 2025