Training load, explained: read the curve without fear

What is training load?

Training load is a number for how much training stress you have taken on, usually duration multiplied by intensity. Heart-rate methods weight each minute by how hard your heart worked; the simplest, session RPE (rating of perceived exertion), multiplies how hard the session felt by how long it lasted.

Session RPE comes from Foster's work on monitoring athletes, which also introduced monotony, the daily average load divided by its standard deviation, and strain, load multiplied by monotony. A later study found session RPE and heart-rate methods highly consistent in their relationship, though absolute scores differed.

Whatever the method, the number means something only against your own history: 300 minutes is a big week if you usually run 120, a quiet one if you usually run 500. The 2017 consensus statement on monitoring athlete training loads treats load as one input among several, read alongside how the athlete responds.

How do fitness and fatigue curves work?

They track two things moving at different speeds: training raises fatigue quickly and fitness slowly, and fatigue fades faster. A curve setting a short recent window against a longer one draws that gap, what you have done lately against what your body has adapted to.

The idea was first formalized as a systems model of training in the 1970s, and most running-app load charts descend from it. It is a simplification: all training stress is interchangeable, and sleep, illness, work and heat do not exist. What it shows well is direction: building, holding or backing off.

Briefly exceeding your usual is how training works. The joint consensus on overtraining describes functional overreaching as a short-term dip in performance that leads to an improvement after recovery. Trouble starts when the imbalance lasts, and the consensus is clear that no single marker meets the criteria to make it generally accepted for telling those states apart. Nor can a curve.

What is the acute:chronic workload ratio, and does it predict injury?

The acute:chronic workload ratio (ACWR) divides your recent load by your longer-term load. It was promoted as a best-practice predictor of training injuries in Gabbett's widely read 2016 paper, but it does not predict injury reliably. Within a few years it drew sustained statistical and conceptual criticism.

Recent load usually sits inside the longer window, so the ratio's two halves are mathematically coupled, which can produce spurious correlation. Impellizzeri and colleagues argued that no study had estimated a causal effect and that the ratio is not consistently related to injury in one direction. In a 2021 reanalysis they divided acute load by made-up chronic values, got similar associations, and found little predictive value over a model with no load information at all. Their conclusion: dismiss the ratio.

Load still matters; one ratio applied to everyone is simply too blunt to predict who gets hurt. Injuries have many causes, most invisible to a watch.

Is the 10% rule right about sudden jumps?

The newest large study points at single runs rather than weekly totals. In the Garmin-RUNSAFE study published in 2025, the week-to-week ratio showed no relationship with injury, while a run more than 10 percent longer than the longest run of the previous 30 days was associated with a higher rate of overuse injury.

The study followed 5,205 runners for 18 months across 588,071 sessions; 1,820 reported a running-related injury, 1,311 of them overuse.

The acute:chronic ratio showed the opposite of the expected pattern: the biggest weekly jumps went with fewer injuries, a hazard rate ratio of 0.75 (0.59–0.96) when weekly load more than doubled. That is an association in observational data, not a reason to double your week. The useful reading is narrower: grow the long run patiently.

How do I read a training load chart?

Read it as this week against your usual. Below it, you are easing off: fine for a taper, a lighter week or a busy patch. Slightly above, you are building. Well above, you are pushing: reasonable for a short block, less so for weeks on end.

• Easing off: less than you usually do. Planned, it is recovery; unplanned for weeks, fitness slowly fades.
• Productive: a little above your usual. Steady building looks like this.
• Pushing: clearly above your usual. Useful in short blocks before a lighter week.
• Too much: well beyond your usual. A cue to make the next runs easy or short, not a diagnosis.
• Recovering: your body is still absorbing recent work, even if load looks modest.

Two habits make any curve more honest. Count everything that loads your heart, long hikes and hard rides included. And read the shape over two or three weeks, not today's dot: one big day moves a short window a lot and means little alone.

When the curve climbs into Pushing or Too much, respond small. Keep the next two or three runs easy and a little shorter, leave the long run where it is, and move a hard session back a day if the last one still sits in your legs. Then look again. Curve settled, sleep and resting heart rate usual: you absorbed the block. Still climbing, easy runs feeling hard: take a properly lighter week.

How often should I take a recovery week?

Regularly, and before you need it. A common pattern is several building weeks, then one with noticeably less running so the work of the previous weeks can sink in. The lighter week is part of the plan, not proof it failed.

Keep easy running easy, too. Foster's monotony measure exists because weeks of identical medium effort pile up strain with little variety in stimulus. Mostly easy days, a little hard work and real rest between hard sessions spread the load so a body can absorb it.

How does Gears handle this?

Under the readiness score, the training load card draws your recent training against your own usual as a curve, with your dot on it, in five zones: Easing off, Productive, Pushing, Too much and Recovering. It counts the heart rate of every workout in Apple Health, from any app and today's included, and how your body is taking it; a run recorded twice counts once. It needs two weeks of runs and carries its own footnote: "A guide to getting the most from training, not a medical reading."

The app's one injury sentence is about the single long run, the pattern the research points to: "Your longest run this week was 40 % longer than any in the 30 days before it. A jump like that is the kind linked with injury: keep the next runs short." The number in it is yours; 40 is an example.

Your personalized training plan follows plain rules: mostly easy running, a lighter week every fourth week ("Lighter week: 25 % less, so the work sinks in."), at most two hard runs a week and never back to back, and no training run more than a tenth past your recent longest.

What it doesn't do: predict injuries, detect overtraining, or attach any injury claim to the weekly zones. It reads only what your devices recorded; strength sessions or a day on your feet that never reached Apple Health are invisible to it.