How to Increase Running Mileage Without Getting Hurt (What the Evidence Actually Says)

Every runner has heard the 10% rule: never increase your weekly mileage by more than 10% a week. It is repeated in training plans, forum posts, and coaching books as if it came down from a mountain. Almost nobody can tell you where it came from, and the one randomized controlled trial that actually tested it, in the most injury-prone group of runners there is, found it did not prevent injuries at all.

That is not an argument for ignoring progression. It is an argument for measuring the right thing. Because the evidence does point to a real risk factor for running injuries, and it is not the size of your weekly total. It is the size of the sudden jump, especially the one long run your body was not ready for.

This is what the trials found, why the spike matters more than the average, how wearable data helps you watch your own progression, and a four-week framework you can run without a coach.

TL;DR: A randomized controlled trial of 532 novice runners preparing for a 4-mile event compared a standard 8-week plan against a graded 13-week plan built on the 10% rule. Injury incidence was 20.8% in the graded group and 20.3% in the standard group: no difference. A one-year cohort study of 874 novice runners found that people who increased their weekly distance by more than 30% over a two-week period had a 1.59x higher rate of distance-related injuries (patellofemoral pain, IT band syndrome, medial tibial stress syndrome, and similar) than those who increased by less than 10%, though the confidence interval crossed 1.0, so it is a trend rather than a proven effect. Meanwhile, the training-load literature shows the opposite of what most runners assume: athletes with a high and consistent chronic training load get injured less, not more, because fitness is protective. The practical conclusion is not "follow a percentage." It is: keep your rolling four-week base steady, cap the size of any single jump, and change only one variable at a time. Wearable metrics such as resting heart rate, heart-rate drift, and training load trends are useful for spotting an accumulating cost early, but they cannot diagnose an injury and they should never overrule pain.

YouTube: Related video

Where the 10% rule came from

The rule is usually traced to running boom coaching advice from the 1970s in the United States, not to any study. Its appeal is obvious: it is a single number, it sounds cautious, and it is easy to follow without knowing anything about how training works.

The problem with a percentage rule is that it ignores everything that actually determines whether a training load is safe for you: your history, your pace, the surface, how fast your long run grows, how much of your running is hard, how much sleep you are getting, and whether you have a cranky tendon from three years ago. Two runners can both "increase 10%" and one of them can be making a very risky change while the other is barely progressing.

There is also a subtle trap in percentage advice. Ten percent of 20 km is 2 km. Ten percent of 60 km is 6 km. The rule asks the most of exactly the runners who have the most tissue stress to manage.

What the evidence actually shows

Three studies are worth knowing because together they replace a folk rule with a testable model.

Read them together and a coherent story appears. Mileage itself is not the villain. Fitness is protective: a runner who has spent months at 50 km a week and adds a hard session is in a different position from a runner who has spent months at 15 km a week and tries the same session. What predicts trouble is the mismatch between what you are asking your body to do now and what it has recently been prepared for. The Buist trial even found that a slower ramp, spread over 13 weeks instead of 8, produced slightly more injuries and noticeably worse adherence, which is a good reminder that "more cautious" is not automatically "safer."

One honest caveat about the Nielsen result: the confidence interval (0.96 to 2.66) crosses 1.0, and the authors call for randomized trials to verify it. It is consistent with coaching experience and with the wider training-load literature, but it is a signal, not proof.

The spike problem: why one long run can matter more than your weekly total

Here is the piece that percentage rules miss entirely. Your injury risk on any given run depends on what that run asked of you relative to what your tissues have recently tolerated. A 25 km long run is not averaged out by the fact that your week totalled only 45 km. It is a single, large, sustained load applied to tendons, bone, and muscle that have never been asked to do it before.

That is why the classic pattern in running injuries is not "someone gradually built up too much" but "someone did one thing that was out of proportion."

• A 12 km long run becomes 22 km because a friend was training for a half marathon.
• A tempo run at 4:45/km pace becomes a 10 km race effort on a Tuesday.
• A flat loop becomes a hilly loop at the same distance and pace, with a 20% higher load for the same training log entry.
• A week of 45 km becomes 70 km because a holiday opened up the calendar.

In each case the weekly total might look reasonable on a spreadsheet, and the tissue load was anything but. This is also why "I only increased by 8%" can still hurt: the increase happened entirely in one session.

Acute versus chronic load, in plain English

Sports science has a useful way to formalize this, described by Tim Gabbett in the 2016 British Journal of Sports Medicine paper. You compare your recent load (the acute period, usually one week) against your medium-term load (the chronic period, usually the last three to four weeks, or your rolling average).

Two numbers matter:

• Your chronic base: what your body is accustomed to. This is your protection.
• The ratio between this week and that base: how big the ask is relative to your preparation.

Worked example. Suppose your last four weeks were 40 km, 42 km, 38 km, and 40 km. Your chronic weekly base is 40 km. Now you plan a 60 km week. That is an acute-to-chronic ratio of 1.5, a 50% jump on your accustomed load. Gabbett's model flags increases like that as the risky pattern, and his headline finding is the counterintuitive part: the well-trained runners with high chronic bases were the ones with fewer injuries.

A 44 km week, by contrast, is a ratio of 1.1. Boring, unremarkable, and the kind of week that builds the engine you will rely on in three months.

You do not need software to do this. Averaging four weeks of mileage, either from a spreadsheet or from your watch's training history, takes two minutes and tells you more than any percentage rule, because it accounts for where you are starting from.

A progression framework that survives real life

The framework below is deliberately not a magic number. It is a set of constraints that leaves room for your own history.

1. Write down your rolling four-week average before you plan anything. That is your baseline, and every progression decision is measured against it.
2. Cap single-week increases at around 10% of the rolling average, and treat 20% as a ceiling you need a reason to break. The reason is not "because I feel great this week."
3. Protect the long run relative to the week. A long run that is more than about a third of your weekly volume is a big share of your load in one session. Most coaching traditions suggest keeping it under roughly 30% to 35% for non-marathon runners, and while that number is convention rather than trial-proven, the underlying idea, capping single-session load, is well supported by the training-load literature.
4. Change one variable at a time. Volume, intensity, frequency, surface, hills, and pace all raise load. Two at once means you cannot tell which one your body objected to, and it doubles the size of the jump.
5. Make every fourth week slightly lighter, not dramatically lighter. A reduction of 20% to 30% in volume every third or fourth week is standard practice. It is not evidence-backed in the same clean way as the load-ratio work, but it costs you almost nothing and it gives you a data point about how you respond.
6. Earn your intensity separately. Adding a hard session to a rising weekly volume is the most common way runners I work with end up on the couch. If your volume is going up, hold intensity steady until the volume feels unremarkable.

What a four-week build looks like in numbers

Two things to notice. The long run never exceeds about a third of the week, and no single step is exciting. The whole point is that a build you can describe as "boring" is one you will complete.

What your wearable can and cannot tell you

Wearable data is genuinely useful during a build, as long as you use it for the right question. The right question is "is the cost accumulating faster than the adaptation?" The wrong question is "am I allowed to run today?"

Useful signals:

• Resting heart rate and overnight HRV trends. A multi-day shift, not a single morning reading, is what matters. Resting heart rate drifting up while HRV drifts down across a week of hard training usually means the load is outpacing recovery. The interpretation pitfalls are well documented, which is why what it means when HRV drops and resting heart rate rises is worth a read before you start overreacting to mornings.
• Heart-rate drift within a session. If your heart rate climbs steadily during a run that used to be steady at the same pace, you are either under-recovered, dehydrated, or running faster than you think. High drift at low intensity usually shows up before a niggle does, and it is one of the clearest signs your aerobic base is not yet matching your planned mileage. Our guide to heart-rate drift in Zone 2 explains how to read it properly.
• Cardio fitness and pace at a given heart rate. Improving over weeks is the adaptation you are buying. Flat or falling while volume rises is a sign you are accumulating fatigue rather than fitness, and training load metrics on Apple Watch are one way to see that trend rather than guessing.
• Sleep duration around your long run. Long-run days that cost you a night of sleep compound into the next week's load. Sleep is where the adaptation actually happens, and if your build is quietly eating your sleep, the numbers on your wrist will reflect that before your calendar does.

Some numbers like the daily readiness-style scores in apps such as Century are best used the same way: as a trend line over one to three weeks, not a green light or a red light for a specific run.

What wearable data cannot do:

• It cannot diagnose an injury. Bone stress injuries, tendinopathy, and most running injuries develop quietly. No HRV reading or readiness score substitutes for pain that changes your gait.
• It cannot see your training history. Your watch knows today's run, not that you took six months off in 2024 or that your left Achilles has been grumpy since a half marathon.
• It cannot tell you what your slowest runs are missing. Runners who go too hard on easy days and too easy on hard days tend to accumulate more fatigue at the same mileage. If your easy runs sit above your aerobic ceiling, your plan needs fixing before your volume does. The talk test is a free way to check that, and how much cardio per week is actually enough is a good sanity check on the overall dose.

Warning signs that mean stop, not "reduce slightly"

Pain rules matter more than any training-load model, because load ratios are population-level tools and pain is your own data.

Back off and get it assessed if any of the following show up:

• Pain that is sharp, localized, or pinpointed to a small area, rather than diffuse and muscular
• Pain that worsens as the run goes on, instead of easing in the first ten minutes
• Any change in your gait, or a limp. This is the single most reliable stop signal in running
• Pain that persists at rest, or that wakes you at night
• Night pain, morning stiffness lasting more than 30 minutes, or a spot on a bone that hurts when you press it. These are classic bone stress injury patterns and they need clinical assessment, not a down week
• A persistent niggle that has not improved after a week of reduced running

The pattern to respect: an injury you catch in week one costs you a week. An injury you train through for three weeks usually costs you two months. The wearables crowd at your wrist is not a diagnostic tool, but it is sometimes the thing that tells you your training is heading downhill before you admit it, which is exactly why it is worth knowing what overtraining signs actually look like rather than only checking your numbers when something already hurts.

The bottom line

The 10% rule is a decent default and a poor law. The trial that tested it head-to-head found it did not prevent injuries in novice runners. What the evidence does support is more useful and more effortful: keep a steady chronic base, cap the size of any single jump, change one variable at a time, and pay attention to whether your recovery signals are tracking your training or falling behind it.

So before your next build, do three things. Look up your rolling four-week average. Write down the week you are planning. Check whether any single run or single jump is doing too much of the work. That is a boring checklist, and it is the version of progression that gets you to the start line in one piece.

Century AI helps you understand your body with a daily health score, recovery score, and sleep insights — using the watch you already wear.