ACWR in ultra-running: how to manage training load and stay off the injury list

The metric that promised to predict injuries before they happen. The science that built it up, then tore it apart. Here's what it actually means for ultramarathon training.

Table of contents

• Who this is for
• Why this matters
• What ACWR is and how it works
• How to measure load: the currency matters
• Why scientists started disagreeing
• What the ultra-running evidence actually shows
• What to monitor instead (or alongside) ACWR
• Overtraining: the continuum that hits ultra-runners hardest
• How to manage training load in practice
• Case study: mountain back-to-back before a 100K
• FAQ
• A flawed metric, a sound principle
• References

Who this is for

This article is for you if:

• You're training for an ultra of 50 km or longer and wondering how to handle rising volume without breaking down.
• You've heard ACWR mentioned by a coach or on a podcast and want to understand what the science actually says.
• You've been injured at peak training before and are looking for a better monitoring system.
• You train without a coach and need practical rules for managing training load on your own.

This article is not for you if:

• You're new to running and looking for a plan for your first 10K.
• You want an app that tracks everything automatically. This is methodology, not software.

Why this matters

Most ultra-runners get injured not because they train too much, but because they did too much too fast. A weekly spike after a few lighter weeks. One hard back-to-back after a month off. Jumping back into full training before the body has recovered from a race. ACWR tried to capture that dynamic in a single number. It didn't fully succeed — but the problem it was trying to solve is completely real.

What ACWR is and how it works

ACWR is a dimensionless ratio built from two values.

Acute load is your training load over the last 7 days. Chronic load is your weekly average over the last 28 days. Divide one by the other and you get a number that tells you whether the past week was normal, easy, or heavy relative to your recent baseline.

• ACWR = 1.0 means you trained exactly as much as your 4-week average.
• ACWR = 1.4 means this week was 40% harder than your norm.
• ACWR = 0.7 means you're backing off significantly.

Example for an ultra-runner

Weeks 1–4: 60, 65, 70, 65 km. Chronic load = 65 km/week. Week 5: 85 km. ACWR = 85/65 = 1.31. Week 5 alternative: 95 km. ACWR = 95/65 = 1.46 — warning zone.

Australian researcher Tim Gabbett proposed a "sweet spot" of 0.8–1.3 based on data from cricket and rugby league. Above ACWR 1.5, injury risk was said to escalate sharply. Below 0.5 — perhaps surprisingly — injury and illness risk also increased. Too little training weakens you.

His most important finding was what he called the training-injury prevention paradox: athletes with high chronic workloads actually sustained fewer injuries, not more. Someone who has been running 90 km per week for months is safer doing 100 km than someone jumping from 40 to 55 km — even though both have an ACWR around 1.1. The chronic load — your accumulated weeks and months of training — is the real protective buffer.

How to measure load: the currency matters

Gabbett's original research used session-RPE: subjective effort rating (0–10 scale) multiplied by session duration in minutes. Simple, cheap, and well-validated by Foster as far back as 1998.

For ultra-runners, several currencies are available and worth combining:

Distance (km) is the simplest but becomes meaningless in mountain terrain without context.

Duration works better in trail running. 5 hours of running is 5 hours of connective tissue loading regardless of how many kilometres the watch recorded.

Km-effort (distance + D+/100) is the metric ITRA uses to classify race difficulty. 30 km with 2,000 m D+ equals 50 km-effort. It accounts for elevation cost and makes sense in technical terrain. Use our flat equivalent calculator to convert mountain sessions to comparable values.

D+ as a separate metric tracks eccentric loading from descents — the primary source of muscle damage in mountain ultras. 2,000 m of descent on hard trail is a fundamentally different stimulus than 2,000 m on soft grass.

Session-RPE captures what no GPS can: heat stress, accumulated fatigue from previous days, poor sleep, altitude, low motivation. The watch doesn't know how you actually felt.

Track both external load (km-effort or duration) and internal load (session-RPE) in parallel. If your RPE for the same distance and terrain rises week over week, fatigue is accumulating. No algorithm will tell you that — but a simple notebook will.

Why scientists started disagreeing

For a few years, ACWR was treated as something close to gospel in sports medicine. Then the cracks appeared.

The mathematical problem

Lolli et al. (2019) noticed that in the standard "coupled" ACWR, the acute week appears in both numerator and denominator — because it's part of the 4-week chronic average. This creates a spurious correlation of approximately r ≈ 0.50 even in randomly generated data. The metric "worked" partly because it was mathematically constructed to look like it worked.

The averaging problem

Menaspà (2017) demonstrated a simpler but important flaw: three runners with completely different weekly training patterns can produce identical ACWR values, because rolling averages can't distinguish how load is distributed. A back-to-back weekend (40 km Saturday, 30 km Sunday, light rest of week) and 70 km spread across 6 days produce the same ACWR — but are fundamentally different stresses on tissue.

The experiment that questioned the ratio itself

Impellizzeri et al. (2021) took real athlete data, calculated ACWR, checked its association with injuries. Then they replaced the real chronic load values with fixed constants and randomly generated numbers. The injury associations were nearly identical. Their conclusion: all the predictive signal in ACWR comes from the acute load — the numerator. The denominator adds nothing.

Their published conclusion:

It is time to dismiss the ACWR as a framework and model.

Gabbett's group responded; critics responded to the response. The debate continues. But the practical implication is clear: ACWR is a tool with real limitations, not a physiological law.

Don't treat the 0.8–1.3 range as a safety guarantee. ACWR 1.2 with poor sleep, badly distributed sessions, and accumulated fatigue is a completely different situation from ACWR 1.2 in a well-rested runner with a strong chronic base.

What the ultra-running evidence actually shows

This is where honesty is required. The direct evidence base for ACWR in ultramarathons consists of exactly two studies. Both from South Africa, both with different results.

Craddock et al. (2020) followed 119 runners over 16 weeks of preparation for the Two Oceans Ultra (56 km). ACWR above 1.5 was significantly associated with injury. ACWR below 0.5 was associated with illness. Running less than 30 km per week increased both risks. High chronic load was protective. This aligns with Gabbett's paradox.

Burgess et al. (2025) tracked 106 runners before and after the Comrades Ultra (87 km). ACWR had minimal influence on injury risk (p = 0.3). Lower overall training loads were associated with higher injury risk during and after the race (p = 0.02).

Two studies, two different answers. You can't build strong conclusions from that.

Frandsen et al. (2025), in a cohort of 5,200 runners, found that single-session spikes were stronger injury predictors than weekly ACWR — and that traditional ACWR even showed a negative dose-response relationship. That is a direct challenge to how the metric is typically used.

Boullosa et al. (2020) articulated perhaps the sharpest critique: in team sports, match-day load is unpredictable — that's where ACWR adds value. In ultra-running, you plan every single session yourself. If ACWR warns you about a spike, it's only because the training plan was poorly designed to begin with. The metric doesn't solve the problem. It reveals that a problem already existed.

Use ACWR as one signal among several, not as the primary decision-maker. Its real value is not risk prediction — it's forcing you to look at load progression with the last 4 weeks in view.

What to monitor instead (or alongside) ACWR

The models are imperfect. Monitoring still matters. Here's what actually holds up.

Session-RPE, training monotony, and strain

Foster (1998) gave endurance athletes a tool that predates ACWR and in some ways outperforms it.

Monotony = mean daily session-RPE ÷ standard deviation of daily session-RPE. It measures how uniform your training is. A value above 2.0 means you're training at similar intensity day after day, without enough variation.

Strain = weekly load × monotony. Foster found that monotony above 2.0 combined with high strain explained 84–89% of illness and injury episodes in his original study.

For ultra-runners, this is particularly relevant. The natural tendency for many of us is to train in the "grey zone" — every session moderately hard, no clear easy days, no genuinely demanding days. Monotony catches that pattern before the body objects.

Calculate your monotony at least once a week. If it exceeds 2.0 during a high-load week, add variation before your body forces a step-back on you.

The CTL/ATL/TSB model

Before ACWR became widely discussed, endurance sport already had a monitoring system — derived from Banister's 1975 fitness-fatigue model.

CTL (Chronic Training Load) is an exponentially weighted average of load over ~42 days. Your fitness.

ATL (Acute Training Load) is a 7-day exponential average. Your current fatigue.

TSB (Training Stress Balance) = CTL − ATL is your form. Positive TSB: rested and capable. Negative TSB: fatigue exceeds fitness. Deeply negative TSB sustained over weeks: the road toward overtraining.

This is essentially ACWR for endurance athletes, but older, better embedded in coaching practice, and calculated automatically by most GPS platforms. Treat it as one signal among several.

HRV as an early warning system

Heart rate variability — specifically the Ln rMSSD parameter measured each morning — reflects the balance of the autonomic nervous system.

Plews et al. (2013) showed that a declining 7-day trend in Ln rMSSD and reduced day-to-day variation signal progression toward non-functional overreaching in elite endurance athletes. Trend matters more than any single measurement, which means consistency is essential — at minimum three times per week, ideally every morning before getting out of bed.

For ultra-runners, HRV captures what kilometres can't: psychological stress, poor sleep, travel to races, altitude, disrupted nutrition. None of that shows up in the training log.

Wellness self-assessment

Halson (2014) found that 84% of high-performance programs use self-report questionnaires as their primary monitoring tool — ahead of any physiological measure or device.

Five questions each morning, scored 1–5: sleep quality, subjective fatigue, muscle soreness, mood, motivation to train. It sounds trivial. But subjective signals often precede objective markers by days, sometimes weeks.

If three sessions in a row have felt like something is off, something probably is — regardless of what the watch says.

Check yourself before checking your watch. Five seconds on sleep and fatigue each morning makes a real difference in how you interpret the day's planned session.

Overtraining: the continuum that hits ultra-runners hardest

The ECSS/ACSM consensus (Meeusen et al., 2013) describes three states on a continuum.

Functional Overreaching (FOR) is planned, controlled overload. Performance drops for a few days. After rest: supercompensation. This is the goal of a well-designed training block.

Non-Functional Overreaching (NFOR) is unplanned overload. Stagnation or declining performance over weeks or months. Symptoms: generalised fatigue, sleep disruption, low mood, reduced motivation, frequent infections. Recovery takes weeks to months.

Overtraining Syndrome (OTS) is extreme, multi-system maladaptation — neuroendocrine, immunological, psychological. It's a diagnosis of exclusion; no single test confirms it. Recovery: months to years.

Kreher & Schwartz (2012) estimated that lifetime incidence of NFOR reaches roughly 60% in elite distance runners. In ultra-running, where training sessions run 4–8 hours and races impose extreme physiological stress, the proportion is likely higher.

A documented case from 2025 (PMC12258013): an ultramarathoner diagnosed with OTS needed approximately 3 years to exceed pre-OTS training benchmarks, returning at 10–15% lower heart rate intensity. Early warning signs that were ignored: hoarse voice after workouts, visibly sunken eyes noted by a partner, disrupted sleep, unstable HRV.

OTS is real. The only effective treatment is prevention.

How to manage training load in practice

Inputs to track

• Weekly external load (km-effort or duration)
• Session-RPE from every training session
• Morning HRV trend (7-day rolling average)
• Daily wellness self-assessment (sleep, fatigue, mood)
• Current CTL/ATL/TSB if using a GPS platform

Rules

• If weekly load spike >30% above your 4-week average → reduce intensity in subsequent days, maintain volume
• If weekly load spike >50% → pull back below threshold, do not continue
• If monotony >2.0 during a high-volume week → add an easy day or rest day
• If RPE rises week over week for the same distance and terrain → take a step-back week regardless of the plan
• If HRV drops more than 2 standard deviations below trend for 3 consecutive days → skip or significantly reduce the planned hard session
• If wellness scores below 2.5 in 3 categories simultaneously → convert planned session to easy or active recovery

Exceptions

A planned overreaching block (a 2-week volume peak, for example) can legitimately break the rules above — but only if it's backed by adequate chronic load and followed by a scheduled recovery week.

Post-race: treat the first 4–6 weeks as a new training cycle. Chronic load has artificially decayed, so the rules above matter more than ever.

Examples

A runner with chronic load of 75 km-effort/week does a 105 km-effort week before a target race (ACWR = 1.40). If they have 6 weeks of consistent building behind them and a good TSB, this is an acceptable spike. The same 105 km-effort week with chronic load at 50 km-effort/week (ACWR = 2.10) — that's a disaster in the making.

Quick reference: load and decision

Weekly monitoring checklist

Warning signs: when to stop without waiting for ACWR

One signal is noise. Several at once, persisting for more than a week, is a call to act.

• Performance declining despite maintained or increased training ("I'm training more but running slower")
• Resting heart rate elevated >5 bpm above personal baseline
• Sleep disruption — difficulty falling asleep or early waking
• Irritability, low mood, loss of motivation — especially before sessions that used to feel rewarding
• DOMS lasting more than 72 hours after a standard training session
• Two or more upper respiratory infections within a month
• Loss of appetite or sudden cravings (a hormonal signal)
• Decreased libido

Case study: mountain back-to-back before a 100K

Scenario: A runner with a chronic load of 65 km-effort/week, 8 weeks out from a 100 km / 5,000 m D+ race. Plans a weekend back-to-back: 45 km Saturday (2,200 m D+) + 30 km Sunday (1,500 m D+). Total weekly km-effort with that weekend: 90 km-effort.

Poor strategy

The runner does this block in a week when chronic load sits at 55 km-effort. Acute/chronic = 1.64. On top of that, Wednesday included a 28 km run to accumulate before peak week. Monotony is high. RPE on Friday is already elevated from Wednesday. The 45-km Saturday runs at 6/10 RPE through the first 20 km, then climbs to 9/10. Sunday's 30 km becomes a quad-soreness march. Monday: lateral knee pain.

Better strategy

The runner builds chronic load to 70–75 km-effort over the preceding 4 weeks. The week before the back-to-back is a step-back week at 40 km-effort. Wednesday before the weekend: easy 12 km. The back-to-back runs at 6/10 RPE Saturday, 7/10 Sunday. Monday: tired but functional. ACWR = 1.25. Adaptation complete.

Use the flat equivalent calculator to convert your mountain sessions to comparable km-effort values and compare weeks across different terrain.

FAQ

Do I need to calculate ACWR to avoid injuries?

No. ACWR is one possible monitoring approach, not the only one and not essential. Tracking weekly load and comparing it to the past 4 weeks gives you similar information. What matters most is tracking something consistently — the specific metric is secondary.

Is the 10% weekly volume increase rule safe to follow?

It has no strong scientific support. Damsted et al. (2018) found no evidence for the rule in a systematic review. Safe progression depends on your current base: at low volume (<40 km/week) you may tolerate 10–20% increases; at high volume (>100 km/week) limit increases to 3–7%.

What should ACWR look like during a taper?

During taper, chronic load starts to drop along with volume. ACWR will fall below 1.0 — this is normal and intended. A low ACWR in taper does not mean you're losing fitness. It means the plan is working.

How do I monitor load in mountains where distance means nothing?

Use km-effort (distance + D+/100) instead of flat kilometres, or track session duration and D+ as two separate metrics. Session-RPE is especially valuable in mountain running because it captures terrain difficulty and fatigue in a way that pace and distance never can.

How long before I can return to normal training after an ultra?

This varies, but a reasonable framework is 4–6 weeks. First two weeks: 30–40% of normal volume. Weeks 3–4: 50–70%. Weeks 5–6: 80–100%. Throughout this period, session-RPE and wellness scores are more important indicators than any number from your watch.

Will an HRV app replace my training load monitoring?

No. HRV is excellent at detecting accumulated fatigue and external stressors. But it won't tell you anything about intensity distribution or single-session spikes. You need both layers.

When should I see a professional instead of managing this myself?

If multiple warning signs are present simultaneously for more than a week and don't improve after a step-back week, that's a signal for a sports physiotherapist or sports medicine physician. Non-functional overreaching that requires months of recovery is real — catching it early is significantly better than waiting.

A flawed metric, a sound principle

ACWR has real mathematical problems. The critics are right on several key points. The evidence base in ultra-running is thin. There is no magic 0.8–1.3 zone that guarantees safety.

But the principle the metric encodes is the most important lesson in training load management the last two decades of sports science have produced:

It is not the load that breaks you. It is the load you are not prepared for.

Build chronic load patiently. Avoid sudden spikes. Monitor more than one variable. Respect step-back weeks. Listen to your body harder than you listen to your watch.

No single number captures the complexity of preparing a human body for 100 kilometres through the mountains. But tracking the right numbers, in context, dramatically improves the odds of arriving at the start line healthy.

Useful tools

• Flat equivalent calculator: convert your mountain training to comparable km-effort and compare weeks across different terrain.
• Training plans: load progression, step-back weeks, and race-specific periodisation built in from day one. You don't need to calculate ACWR; the methodology handles it.

References