The long-run rule that actually prevents injuries
Keep every run within 10% of the longest run you have completed in the last 30 days. In a study of 5,205 runners, a single run more than 10% longer was linked to a 52–128% higher rate of overuse injury. Week-to-week mileage increases showed no such link.
What to do
- Before a long run, find your longest run of the past 30 days and multiply it by 1.1. That is your ceiling.
- Grow your long run in steps of up to 10%, and only from runs you have actually completed, not runs you planned.
- After illness, injury or a holiday, the ceiling drops with your recent longest run. Rebuild from there instead of jumping back to your old long run.
- If you are new to running or were injured in the last year, consider smaller steps, around 5%. That is our cautious suggestion, not a tested rule.
- Stop worrying about the weekly 10% rule. It has not held up in trials.
The study
The Garmin-RUNSAFE study followed 5,205 runners who trained with Garmin watches, with an average age of 46 (Frandsen and colleagues, 2025). The researchers had data from 588,071 training sessions.
They compared each run with the longest run the same person had done in the previous 30 days. When a single run was more than 10% longer, the rate of overuse injury went up:
| How much longer than the longest run in the past 30 days | Change in injury rate |
|---|---|
| Up to 10% longer | No clear increase |
| 10–30% longer | About 64% higher |
| 30–100% longer | About 52% higher |
| More than double | About 128% higher |
At the same time, week-to-week changes in total distance showed no link with injury. In this data, the risk came from single runs that were long compared with recent experience, not from busy weeks.
What a spike looks like
The rule is easy to break without noticing. If your longest run in the last month is 21 km, a 26 km run is a 24% jump and counts as a spike. The ceiling for that next run is 23 km.
Building a long run from a half marathon to marathon-training distances in 10% steps looks like this:
| Step | Longest completed run | Ceiling for the next run |
|---|---|---|
| Start | 21 km | 23 km |
| 1 | 23 km | 25 km |
| 2 | 25 km | 27.5 km |
| 3 | 27.5 km | 30 km |
| 4 | 30 km | 33 km |
That is four or five steps from 21 km to 32 km. Many plans alternate a longer run with a shorter one, so the build takes 8–10 weeks. If a plan asks for a bigger jump, either start the build earlier or accept a shorter longest run. The ceiling counts every run, so the same logic applies to a long midweek run or a trail race.
Why the weekly 10% rule doesn’t hold up
The familiar rule is “don’t increase weekly mileage by more than 10%”. It has been tested directly:
- In a randomized trial of 532 novice runners, a programme that followed the 10% rule had the same injury rate as a faster one, 20.8% against 20.3% (Buist and colleagues, 2008).
- In 874 novice runners, weekly increases above 30% showed only a weak, non-significant trend towards more injuries (Nielsen and colleagues, 2014).
Another popular measure, the acute:chronic workload ratio (ACWR), compares the last week’s load with the last month’s. It has been heavily criticized, because the way it is calculated can produce links with injury even in simulated data where none exist (Impellizzeri and colleagues, 2020). In the Garmin data, a higher ACWR was actually linked to fewer injuries.
The sensible reading: sudden jumps matter, and comparing each run with your recent longest run catches those jumps more directly than weekly totals do.
Who should be most careful
Running injuries are common. Around 40% of runners report an injury during a training period (Kakouris and colleagues, 2021). Some runners are at higher risk:
- New runners get injured more than twice as often per hour of running as experienced recreational runners (17.8 against 7.7 injuries per 1,000 hours; Videbæk and colleagues, 2015).
- A previous running injury is the strongest known risk factor (van Poppel and colleagues, 2021).
For these groups, smaller steps than 10% are a reasonable precaution. No study has tested a specific smaller number.
What we don’t know
- It is observational. The study shows a link between spikes and injury, not proof that spikes cause them. Still, a sudden long run is a plausible cause.
- The runners were Garmin users with an average age of 46. Results may differ for younger, faster or brand-new runners.
- The 10% line is where risk rose in this data, not a sharp biological limit. A 9% jump is not safe and an 11% jump is not dangerous. It is a practical ceiling.
- It doesn’t cover everything. Speed, hills, surfaces, shoes and recovery also matter and are not captured by run length alone.
Sources
- Frandsen et al. (2025). Garmin-RUNSAFE: single-session running distance spikes and injury in 5,205 runners. British Journal of Sports Medicine.
- Buist et al. (2008). Randomized trial of a graded training programme (the 10% rule) in novice runners.
- Nielsen et al. (2014). Weekly progression in running distance and injury in 874 novice runners.
- Impellizzeri et al. (2020). Critique of the acute:chronic workload ratio.
- Kakouris et al. (2021). Systematic review of running-related injury incidence.
- Videbæk et al. (2015). Injury incidence per 1,000 hours of running in different runner groups.
- van Poppel et al. (2021). Risk factors for running injuries.
Last reviewed 12 November 2026.