Recovery & Injury Running · · 8 min read

The 10% Weekly Mileage Rule Is Wrong — What 23,047 Runners Actually Show

The 10% rule has no founding research paper. A 532-person RCT found zero benefit. Here's what the real injury data says you should do instead.

AO
AthleteOS Data Science
TL;DR — The Answer

The 10% weekly mileage rule has no peer-reviewed origin — it's coaching folklore. A 2008 RCT (n=532) found injury rates of 20.8% vs. 20.3% with and without it (p=0.90). The real injury signal is single-session spikes: one run >110% of your longest run in the prior 30 days raises overuse injury risk by 64%. The evidence-based target is an ACWR of 0.8–1.3, calculated using an exponentially weighted moving average.

Nobody can tell you where the 10% rule came from. Sports scientists have searched. No founding paper exists. The rule spread through coaching manuals in the late 1980s and early 1990s — not through research. And in 2008, a 532-person randomized controlled trial tested it directly and found it did absolutely nothing to prevent injury.

Yet it’s still the first piece of advice every new runner hears.

The 10% Rule Has No Science Behind It

The 10% rule — the idea that you should never increase your weekly mileage by more than 10% — is officially classified as “injury-prevention folklore” by researchers who have tried to trace its origin. The 2018 systematic review by Damsted et al. in the International Journal of Sports Physical Therapy put it plainly: no evidence exists for its use.

The single most damning piece of data comes from Buist et al. (2008). Researchers randomized 532 novice runners into two groups. One followed a 13-week graded program built around the 10% rule. The other followed a faster 8-week standard program. Injury incidence was 20.8% in the 10% rule group vs. 20.3% in the standard group (p = 0.90). Statistically identical. Zero benefit from the slower ramp.

That’s the only RCT ever run on this rule. It failed.

The Real Injury Data — What 23,047 Runners Actually Show

The most comprehensive review of running injury and training load comes from Fredette et al. (2022), a systematic review of 36 studies covering 23,047 runners. Their conclusion was stark: no universal training progression recommendation can be issued based on current evidence.

What they did find is a pattern. Injured runners averaged 31.6% weekly progression before getting hurt. Uninjured runners averaged 22.1%. In the week before injury, the spike was even sharper — injured runners showed 86% greater progression than uninjured ones (p = 0.026).

A threshold closer to 30% is where risk rises, not 10%.

That gap matters. Think of your body’s connective tissue like a suspension bridge cable. Each week of steady loading sends a slow, controlled signal to strengthen the cable. A sudden large load doesn’t send a stronger signal — it just snaps the wire before it has time to adapt. The 10% rule treats all cables as equally fragile. They aren’t.

ACWR Zones and Injury Risk Under-training (<0.8) Under-training Sweet spot (0.8–1.3) Lowest risk Caution (1.3–1.5) Elevated Danger zone (1.5–2.0) High risk Very high risk (>2.0) Danger Gabbett 2016 BJSM framework. Bar height represents relative injury risk, not a precise injury rate — the sweet spot band (0.8–1.3) produces the fewest injuries across team sports and running cohorts.

What the Weekly Ramp Rule Gets Wrong

The 10% rule treats every runner as identical. The data says the opposite.

Videbæk et al. (2015) pooled 13 studies and found that novice runners sustain 17.8 injuries per 1000 hours of training. Recreational runners average just 7.7 per 1000 hours. Ultra-marathon runners: 7.2 per 1000 hours.

That’s a 2.3x difference. Same rule, two completely different risk profiles.

There’s a second problem. At 10 miles per week, a 10% increase is 1 mile. At 60 miles per week, it’s 6 miles. The same percentage produces 6x the absolute stress. The rule is mathematically incoherent across fitness levels.

Under-ramping has a cost too. Rasmussen et al. (2013) followed 662 marathon finishers and found that those training below 30 km per week before race day had twice the injury risk of the 30–60 km group (RR = 2.02, p < 0.01). Being too conservative isn’t safe — it just shifts the risk from overuse to under-preparation.

Session Spikes Beat Weekly Ramps as the Real Injury Signal

Here’s what changes everything. A 2025 British Journal of Sports Medicine study tracked 5,205 runners over 18 months, logging 588,071 individual sessions via GPS. The researchers tested both week-to-week volume changes and per-session spikes against injury outcomes.

Week-to-week load change: non-significant predictor of injury.

Single-session spikes: dominant injury signal.

A single run exceeding 110% of your longest run in the prior 30 days raised overuse injury risk by 64% (HRR = 1.64, 95% CI: 1.31–2.05). A run exceeding 200% of that threshold doubled the risk (HRR = 2.28). The week didn’t matter. The individual session did.

That’s a different model entirely. The danger isn’t slow weekly creep — it’s the one big day.

ACWR Over a 12-Week Marathon Build (Illustrative) 1 1 1 2 2 ACWR Wk1Wk2Wk3Wk4Wk5Wk6Wk7Wk8Wk9Wk10Wk11Wk12 Controlled build (sweet spot) Injury-risk pattern
Stylized illustration of a controlled build vs. a spike-heavy pattern. The dashed danger zone starts at ACWR 1.5. The controlled athlete stays in the 0.8–1.3 sweet spot most weeks and uses a recovery week at weeks 4, 8, and 12.

The Better Framework: ACWR and Your Fitness Score

The evidence-based replacement for the 10% rule is the acute-to-chronic workload ratio (ACWR). It compares your last 7 days of training load against your rolling 28-day average.

ACWR = 7-day load / 28-day average load

A ratio between 0.8 and 1.3 is the sweet spot. Above 1.5 is the danger zone. Above 2.0, studies show very high injury risk — English Premier League data found a relative risk of 6.7 for non-contact injuries at low chronic load combined with high acute load.

But the calculation method matters. Murray et al. (2017) compared simple rolling-average ACWR against an exponentially weighted moving average (EWMA) in 59 elite athletes over 2 years. The EWMA model was significantly more accurate at detecting injury risk (p = 0.001 in the very-high-ACWR zone). EWMA gives more weight to recent days, which better mirrors how your body actually recovers.

The difference: simple rolling averages treat Monday and Sunday of last week equally. Your body doesn’t.

Most consumer platforms — including TrainingPeaks and WHOOP — use EWMA under the hood. The 10% rule uses nothing.

A Better Benchmark Table

Runner profileCurrent weekly mileage10% rule capACWR-safe cap (approx.)Risk if under-ramping
Absolute beginner10 mi+1 mi+2–3 miHigh — connective tissue adapts slowly but needs minimum stimulus
Recreational runner25 mi+2.5 mi+5–7 miModerate — under 30 km/wk pre-marathon = 2x injury risk
Marathon build45 mi+4.5 mi+7–10 miLow — fitness base handles larger absolute adds
Experienced runner60 mi+6 mi+10–15 miVery low — high CTL tolerates aggressive weekly gains

The 10% column produces meaningless ramp increments at low mileage and overly conservative ones at high mileage. The ACWR column adapts to the runner.

Meet the Rule in Practice: A Case Study

James is 42, running his third marathon, currently at 38 miles per week. He decided to jump from his usual Saturday long run of 13 miles to 18 miles in a single week. That one session sat at 138% of his longest run in the prior 30 days — well past the 110% spike threshold the 2025 BJSM study identified.

His weekly mileage increase was only 13%. Under the 10% rule, he was already “over” the cap. Under the ACWR model, the red flag was the single session, not the week total.

He strained his left calf the following Tuesday. Three weeks off.

His training partner, Elena, runs the same mileage. She increased her long run from 13 to 14 miles the same week — 108% of prior 30-day max, just under the spike threshold. Her week-to-week increase was about 6%. Her ACWR hit 1.18. She ran the marathon eight weeks later in 3:41.

One session made the difference. Not the weekly percentage.

How AthleteOS Handles Your Mileage Ramp

AthleteOS doesn’t apply a blanket 10% cap. The AI coach tracks your fitness score (CTL) and calculates your ACWR using an exponentially weighted moving average — the Murray et al. approach that outperforms rolling averages in injury prediction. Your weekly mileage cap adjusts to your current fitness, not a fixed percentage.

It also flags when any single session is set to exceed 110% of your longest run in the prior 30 days. That’s the session-spike trigger the 2025 Garmin-RUNSAFE study identified as the primary overuse injury driver.

For a complete picture of how training load connects to injury risk, see how Zone 2 builds your aerobic base without accumulating injury stress and how aerobic decoupling signals when your base is ready for more volume. If you’re returning from time off, the return-to-run framework covers how to restart your ACWR safely from zero.

Set up your running training plan in AthleteOS to get your ACWR tracked automatically from day one.

The 10% rule sounds sensible. The research doesn’t support it. Your long run on Saturday is far more important than what percentage it represents of your week.

Watch the session. Not the cap.

Frequently Asked Questions

Is the 10% rule backed by science?

No. Sports scientists who searched for its origin found no founding peer-reviewed paper. The only RCT to test it directly found identical injury rates in the 10% group and a faster-ramp control group (20.8% vs. 20.3%, p=0.90).

What is a safe weekly mileage increase for runners?

Research suggests staying below 30% over any 2-week window. The evidence-based target is an ACWR between 0.8 and 1.3. More importantly, avoid single-session spikes above 110% of your longest run in the prior 30 days.

What is ACWR and how is it calculated?

ACWR is the acute-to-chronic workload ratio — your recent 7-day load divided by your rolling 28-day average. An ACWR of 0.8–1.3 is the sweet spot. Above 1.5 enters a danger zone. The most accurate version uses an exponentially weighted moving average rather than a simple rolling average.

Can under-training increase injury risk?

Yes. Runners averaging less than 30 km per week before a marathon had 2x the injury risk of those in the 30–60 km range (RR=2.02, p<0.01). Staying too conservative has its own cost.

Do novice runners need different ramp rules than experienced runners?

Definitely. Novice runners sustain 17.8 injuries per 1000 hours of training. Recreational runners average 7.7 per 1000 hours. A single percentage cap can't serve both populations.

What is a good ACWR for marathon training?

Aim to keep your ACWR between 0.8 and 1.3 throughout your build. Don't let it drop below 0.8 (under-training zone) or climb above 1.5. Build slowly enough that you spend most weeks between 1.0 and 1.2.

#running-injury#mileage-ramp#ACWR#training-load#injury-prevention

Stop guessing your safe mileage cap

AthleteOS tracks your ACWR with an exponentially weighted moving average and flags session spikes before you run them. Your ramp rate is based on your fitness, not a number from a 1980s coaching manual.

Generate Your Free AI Plan
14-day free trial · No credit card required