Not All Fatigue Is The Same

PerformanceSprintsRecovery
Written By Cat Hoh

A sprinter fading in the last 30m of a 400, a marathoner whose stride falls apart at the 35km mark, and a trail runner losing three minutes on a descent they climbed easily in training, are not experiencing the same thing. They're failing in three different ways, in three different tissues, for three different reasons. Coach all three of them like "fatigue" is one problem and you'll fix none of them well.

Twenty-nine studies, mostly systematic reviews and meta-analyses from the last ten years, say the same thing from different angles: the type of muscle contraction involved, concentric, eccentric or isometric, decides how fast an athlete fatigues, what actually gives out first, and how long it takes to come back. That's true whether the event is on a track, a road, or the side of a mountain.

Eccentric isn't automatically the fatiguing one

The standard assumption is that lengthening contractions (eccentric) are the more fatiguing, more damaging mode and shortening contractions (concentric) are the resilient workhorse. In the legs specifically, a lot of the evidence says the opposite.

Figure 01 / Interactive

The same fatigue test, opposite result depending on the muscle group

Body region / tap to change

Upper body Lower body

Concentric strength loss 33.6% average

Eccentric strength loss 31.4% average

In the arms and shoulders, concentric and eccentric fatigue come out roughly equal. Neither mode has a clear advantage.

Concentric strength loss 39.7% average

Eccentric strength loss 13.3% average

In the legs, it flips. Eccentric strength loss is a third of the concentric loss. The muscles that absorb load on every stride are comparatively resistant to fatiguing that way.

Data. Pooled averages (±SD not shown) from a systematic review of roughly 30 acute strength-loss studies: Nuzzo, Pinto & Nosaka (2023), Scandinavian Journal of Medicine & Science in Sports 33(10):1901-1915. The authors attribute the lower-body difference to muscle architecture and habitual loading, not a universal neural rule.

What eccentric contractions do lose disproportionately isn't force, it's control. Motor units fire more slowly during lengthening than shortening at a matched force, and the brain struggles to fully drive a muscle while it's lengthening: voluntary activation deficit during a maximal eccentric quad contraction averages around 21%, against only 7 to 8% for an isometric contraction of the same muscle (Duchateau & Enoka 2016). That's a genuine central-nervous-system limit, not muscle fatigue in the usual sense.

The damage side runs on its own clock, and it isn't caused by lengthening itself. It's caused by individual sarcomeres being forced past their optimal length unevenly, which shears the contractile filaments and disrupts the Z-band, letting calcium flood in and triggering an enzyme (calpain) that breaks down structural protein. Damage tracks with tension and speed, not with "eccentric versus concentric" as a category (Paschalis et al. 2026). The resulting recovery schedule is fairly consistent: torque deficit runs 2 to 7 days, soreness peaks at 24 to 72 hours and clears by day 7 to 10, and creatine kinase peaks at 24 to 48 hours and normalises by around day 5.

Running and every jump event add a third layer on top of this: the stretch-shortening cycle, a rapid eccentric pre-load immediately followed by a concentric push, with the tendon doing some of the work in between. It fatigues on its own timeline, distinct from either contraction mode alone, and that timeline is longer and messier than either: elite athletes tracked through a hard training block took 4 to 8 days to recover stretch-shortening function, with wide individual variation (Philipp et al. 2024). Prior exposure to the same load blunts it next time, the repeated bout effect showing up here exactly as it does in soreness (Kamandulis et al. 2019).

Track and field: the fatigue is mostly in the muscle, not the head

Repeated-sprint fatigue is usually explained through phosphocreatine depletion, metabolite build-up and impaired anaerobic and oxidative metabolism, with a neural contribution layered on top (Girard, Mendez-Villanueva & Bishop 2011). A controlled 12 x 40m protocol on 30-second recoveries put numbers on the split: sprint time rose 17% from first rep to last, plantar flexor strength fell 11%, twitch peak torque fell 13%, tetanic torque fell 17%. Voluntary activation, the neural piece, dropped only 2.7%. The authors' read: this kind of fatigue is overwhelmingly peripheral, happening in the muscle itself rather than a drop in neural drive, and it recovers fully within 30 minutes (Perrey et al. 2010).

Reactive strength index (RSI), which captures how efficiently an athlete uses the stretch-shortening cycle on ground contact, correlates with 0 to 30m acceleration, 30 to 100m top speed and change-of-direction speed across the pooled literature (Jarvis et al. 2022). A horizontal, running-specific RSI test tracks sprint times even more tightly than the standard vertical drop-jump version, and lines up with muscle-quality measures in the vastus lateralis (Ciacci, Nigro & Bartolomei 2024). Neither paper measured RSI decline within a session directly, but together they establish RSI as the exact thing stretch-shortening fatigue erodes when a sprinter or jumper looks flat late in a session.

It's genuinely split. In 210 Australian footballers, low eccentric hamstring force on the Nordic curl was linked to 2.7 to 4.3 times the injury risk (Opar et al. 2015), and that finding shaped a lot of programming. A 2021 meta-analysis then pooled six cohort studies and over 1,100 athletes and found no significant difference in eccentric hamstring strength, in any measure, between athletes who went on to get injured and those who didn't (Opar et al. 2021).

The exercise itself is still well justified: an umbrella review found Nordic hamstring training cuts strain injury by up to 51%, tied to fascicle-length gains and how consistently athletes actually do it (Ferreira et al. 2024). Keep the exercise. Be more careful about reading one low score as a red flag for one athlete.

Middle distance is the one gap in this whole brief. The most current review of world-class 800m and 1500m training treats fatigue resistance as decisive for performance but doesn't break it down by contraction mode at all (Tønnessen et al. 2021). Anything applied to that event group here is extrapolated from the sprint and stretch-shortening literature above, not directly demonstrated.

Road running: strength loss that scales with distance

The longer the effort, the more contraction capacity it costs, and eccentric capacity often goes first.

Figure 02 / Interactive

Drag through 5km, a half marathon and a marathon

Race distance / drag to change

5kmHalf marathonMarathon

Quadriceps strength loss ~15%

A 5km effort costs real strength, but the authors put it down to metabolic and neural fatigue, not structural damage. There's not enough distance or eccentric loading yet for that.

Eccentric hamstring torque loss 19.8%

Concentric knee-flexor torque loss 10.1%

Eccentric hamstring capacity falls almost twice as far as concentric. The eccentric-to-concentric strength ratio drops from 0.78 to 0.68, a pattern linked elsewhere to hamstring and ACL injury risk.

Isometric knee extensor MVC loss 22%

Plantarflexor MVC loss 17%

Still down five days later. The authors traced this to central, not peripheral, fatigue mechanisms.

Cross-study synthesis, not one continuous curve. 5km data from Findrik, Šušnjara & Kuna (2026), Sports 14(7):262; half marathon from Wang et al. (2025), Sports Medicine – Open 11(1):22; marathon from Petersen et al. (2007), European Journal of Applied Physiology 101:385-396. Three separate protocols and cohorts, not one tracked group, so treat the progression as an order of magnitude, not a precise curve.

The mechanics degrade on a schedule of their own, and mostly earlier than you'd expect. Across marathon distance, ground contact time rises about 7% within the first hour then plateaus; stride length falls from roughly 2.04m at 8km to 1.41m at 40km, and the steepest single drop, 13%, happens in that first hour rather than at the end (Olaya-Cuartero et al. 2024). A field study using wearable sensors found the same thing from another angle: the proportion of the stride spent in ground contact starts rising around the 34km mark, and how much it drifts correlates directly with the gap opening up between heart rate and pace, the mechanical signature of the fade every marathoner recognises (Hunter, Lena & Muniz-Pumares 2025). A broader review across running distances found the same direction of travel every time it looked: contact time, flight time and tibial shock go up, peak ground reaction force and leg stiffness go down (Apte et al. 2021).

Any road course with real descents borrows straight from the trail physiology below: reported knee extensor and plantarflexor strength losses of 14 to 55% after downhill-loaded running come from eccentric quadriceps demand on the way down (Lu et al. 2025), and that same mechanism is named as a contributor to patellofemoral pain and plantar fasciitis risk in distance runners generally. No review directly compares marathon, half marathon and 5 to 10km races on central-versus-peripheral fatigue; the standing view is that both contribute, and the balance shifts with duration rather than following a fixed rule (Millet & Lepers 2004).

Trail and mountain ultra-running: the descent is the whole story

Take the same fatigue physiology and add sustained descents, and the eccentric cost stops being one factor among several. It becomes the dominant one.

Figure 03 / Interactive

Where the energy cost of running actually goes up

Terrain / tap to change

Level Uphill Downhill

Change in energy cost after a mountain ultra +3.9%, not significant

Level running barely costs more after the race than before it. Fatigue hasn't really touched this mode.

Change in energy cost after a mountain ultra +2.1%, not significant

Uphill running is essentially unaffected too. The concentric-dominant climb holds up.

Change in energy cost after a mountain ultra +13.1%, p<0.001

Downhill running is where efficiency actually collapses. Almost the entire fatigue cost of the race concentrates in the eccentric-loaded mode.

Data. Energy cost measured before and after a mountain ultramarathon, split by terrain: Vernillo et al. (2015), Journal of Sports Sciences 33(19):1998-2005. Bar lengths are scaled for visibility, not a shared percentage axis.

A single 30-minute bout at a -20% gradient is enough to drop knee extensor strength 25%, peak muscle activation 22%, and the late phase of rate of force development 25%, while the early phase is untouched, pointing to a specific problem with contractile tissue and tendon stiffness rather than a blanket strength loss (Varesco et al. 2022). At 45 minutes, downhill running produces measurably more torque loss and peripheral fatigue than uphill or level running at matched intensity, with strength losses of 14 to 55% in the quads and 15 to 25% in the calves, taking 4 to 5 days to fully resolve (Bontemps et al. 2020).

24.46%

Bigger, the gap between downhill and uphill speed in the second half of a trail ultramarathon compared with the first half. Even elite finishers aren't exempt: they slow 15.87% (men) and 20.04% (women) on descents between the first and second half of a race.

Source: Genitrini et al. (2022), Journal of Functional Morphology and Kinesiology 7(4):103.

The damage at ultra distance is large but rarely dangerous. After the Tor des Géants, 330km with roughly 24,000m of both climbing and descending, knee extensor strength was down 23.9% at the finish and creatine kinase averaged 3,719 IU/L (Saugy et al. 2013). Pooled across ultra-trail races generally, CK rises by a median of roughly 5,370% pre-to-post, and yet clinical rhabdomyolysis was diagnosed in only 0.28% of runners studied, despite likely under-diagnosis (Lecina et al. 2024). A separate review put average post-race knee extensor strength loss at around 18% (García-Valiente et al. 2026). Downhill running produces more damage than uphill, but total race elevation doesn't predict CK in a straight line, so the course profile matters more than the total climbing.

The best-evidenced protection is simply prior exposure: runners already accustomed to downhill running show markedly less damage on the same descent than unaccustomed runners, the repeated bout effect again (Bontemps et al. 2020). There's also direct evidence for high-pressure compression garments worn during a downhill bout: they cut the acute strength loss from 18.2% to 13.9%, and roughly quartered the residual deficit at 24 hours, from 10.4% down to 2.6% (Ehrström et al. 2018).

Quick glossary

Tap a card

What this means for programming

  • Don't assume eccentric work is the more fatiguing mode. In the legs specifically, it's often the more resilient one under fatigue. What degrades first is fine control and force accuracy, not raw output.
  • Downhill exposure is the biggest lever you have in any event with a descent, from a single training run to a 330km race. Give the legs prior exposure well before race day, not the week before.
  • Mechanics matter as much as strength as a race or session goes on. Ground contact time and leg stiffness change before an athlete reports feeling weak, and they lead the pace drop that follows.
  • Use a low Nordic hamstring score cautiously as an individual injury flag, the evidence for that specific use is weak, but keep Nordic training in the programme regardless: the exercise has strong injury-reduction evidence independent of any single test score.
  • For anything longer than a half marathon with real descents, the eccentric and downhill cost dominates the fatigue picture, not the aerobic cost. Programming eccentric strength and downhill-specific exposure is the main variable separating who slows down on descents late in a race and who doesn't.

Hold this loosely at the edges. Middle-distance track events have no dedicated contraction-mode fatigue literature at all, everything applied to them here is extrapolated from sprint and stretch-shortening research. The eccentric hamstring injury-prediction evidence is genuinely mixed, and both sides are presented above rather than the more convenient one. The road-running distance progression in Figure 02 is a synthesis across three separate studies and protocols, not one tracked group, so treat it as an order of magnitude rather than a validated curve. Most trail and ultra data comes from large-elevation, Alps-style courses, how well it transfers to flatter or shorter Australian trail formats isn't directly established.

References

All links checked September 2026.

Cat Hoh
Next

Same Speed. Different Shapes.