TriForward
Bike ·

Why Watts Fade: It's Not Just an Empty Fuel Tank

The mechanisms behind late-ride power loss, honestly graded: fibre-level glycogen depletion, early central fatigue, rising oxygen cost, and a failed hunt for a single muscle marker.

Why Watts Fade: It's Not Just an Empty Fuel Tank

Sprint Summary

The short version — read this if you're short on time.

The empty-tank story about late-ride power loss is intuitive, but the strongest direct test of it — correlating muscle glycogen change with power-duration change — found only a weak relationship. The better-supported current picture is a mix of fibre-level glycogen depletion in your most oxidative muscle fibres, a nervous system that backs off before the muscle itself shows measurable transmission failure, and a genuine, still not fully explained, loss of pedalling efficiency. Carbohydrate feeding helps, meaningfully, but doesn't eliminate any of this. The honest grade for this whole area is mixed: moderate evidence for the phenomena, limited evidence for the causes. That's not a satisfying answer, but it's the accurate one.

Safety & Context

Most mechanistic studies referenced here used trained or elite male cyclists in controlled, sometimes fasted or water-only lab protocols; do not attempt to replicate fasted or water-only long rides as a personal experiment. Always fuel long rides according to current carbohydrate-intake guidance, and consult a sports dietitian if you have gut, metabolic or cardiovascular conditions before changing your fuelling strategy.

Full Distance

The complete research and analysis.

Ask most riders why their power falls apart late in a long ride and you'll get the same answer: they ran out of fuel. It's an intuitive story, it lines up with how bonking feels, and it's the reason the sports-nutrition industry exists. It's also, according to the strongest test of it we could find, the mechanism the data support least directly. That doesn't mean fuelling doesn't matter — it clearly does. It means the honest picture of why watts fade is more complicated than an empty tank, and worth understanding if you want to know what you're actually training when you chase durability.

This article grades every mechanism it covers as either moderate, limited, or established-but-not-fully-explained. Read it as an honest map of what's known, not a tidy story — because a tidy story isn't what the evidence currently supports.

Why this matters for your training, not just your trivia

If you assume power loss is purely a fuelling problem, you'll fix the wrong thing. You might chase ever-higher carbohydrate intakes expecting them to eliminate late-ride fade entirely, get frustrated when they don't, and miss that some of what's happening to your legs late in a ride has nothing to do with what's in your gut. Knowing which mechanisms respond to fuelling and which don't tells you where your actual leverage is.

The mechanism everyone assumes: running out of glycogen

Start with what's genuinely well established. Feeding carbohydrate during prolonged exercise reliably slows the loss of critical power, in a clear dose-dependent way. In 16 endurance-trained cyclists and triathletes on a standardised carbohydrate-loading protocol, critical power fell in every condition after 3 hours of riding (p < 0.001) — but by different amounts: 236 ± 30 W on water only, 257 ± 28 W at 60 g/hour, and 266 ± 29 W at 120 g/hour, against 277 ± 27 W fresh, with 120 g/hour beating 60 g/hour beating water (Norte, Slinn, Johnson, Mahon, Shepherd, Strauss & Louis, Scandinavian Journal of Medicine & Science in Sports, 2026). Earlier work found the same shape: 60 g/hour negated what would otherwise have been a 9% fall in end-test power after 2 hours of heavy-intensity cycling (Clark et al., Journal of Applied Physiology, 2019). This part of the story is strong and worth acting on: carbohydrate during a long ride buys you real, measurable watts.

Here's where the story gets uncomfortable for the empty-tank narrative. When researchers actually measured muscle glycogen directly, with biopsies before and after 2 hours of heavy-intensity cycling, the change in muscle glycogen barely tracked the change in the power-duration relationship at all: ["the change in muscle [glycogen] was not significantly correlated with the changes in either" end-power (r = 0.19) "or" W′-equivalent (r = 0.07)](https://pubmed.ncbi.nlm.nih.gov/30995104/), and the authors state outright that the changes in those performance parameters "are not primarily determined by changes in muscle [glycogen]." That is about as direct a contradiction of the empty-tank story as this literature offers: whole-muscle glycogen level and the power you can produce are, on this evidence, only weakly related.

Fuelling well doesn't make the problem disappear either. Even with an aggressive 100 g/hour of carbohydrate across a 4-hour intermittent protocol, 12 elite and professional cyclists still lost 10% of their 6-minute time-trial power and 6% of peak power (Ørtenblad, Zachariassen, Nielsen & Gejl, European Journal of Applied Physiology, 2024). Fuelling is the single biggest lever an age-grouper has over late-ride power — but it's a partial fix, not a cure, and the mechanism it's fixing isn't simply "the tank going empty."

A better-resolution picture: it's not the whole tank, it's specific fibres

If whole-muscle glycogen doesn't explain the power loss, something at a finer resolution might. Single-fibre electron microscopy work gives the best account currently available. Researchers found that the muscle fibres which remain non-glycogen-depleted during prolonged exercise are systematically the ones with less mitochondria: mean mitochondrial volume as a share of fibre volume in the non-depleted fibres was measured at 2% lower after 1 hour, 5% lower after 2 hours, and 12% lower at the point of task failure, compared with all sampled fibres (Nielsen, Jensen & Ørtenblad, Scandinavian Journal of Medicine & Science in Sports, 2024). The authors' interpretation: a glycogen-dependent fatigue of individual fibres during submaximal exercise may reduce the muscle's overall oxidative power.

Read that carefully, though, because the grading matters. The confidence intervals around those numbers are wide — the 1-hour and 2-hour estimates both include zero — and only the task-failure timepoint clears that bar. This is a genuinely plausible mechanism, and it's the best current account of why sustainable power can fall without your legs literally running empty: your most metabolically capable fibres burn through their local glycogen first and effectively drop out of the rotation, leaving the work to less capable fibres that cost you more oxygen for the same power. But it's mechanistically moderate at best, and limited as a full explanation of the performance loss — it's one piece, not the whole picture.

Your brain backs off before your muscle actually fails

A separate, well-replicated thread concerns the nervous system rather than the muscle itself, and it has a clear, useful time course. After 5 hours of cycling at a moderate intensity in 9 male subjects, maximal voluntary contraction fell around 18% and voluntary activation — a measure of how much of the muscle the brain is actually able to switch on — fell about 8%. Changes in the M-wave, an index of how well electrical signals travel across the muscle membrane once they arrive, showed up only after 4 hours (Lepers, Maffiuletti, Rochette, Brugniaux & Millet, Journal of Applied Physiology, 2002). In plain terms: your brain's willingness to drive the muscle hard falls before the muscle's own wiring shows measurable trouble. Central fatigue arrives first; peripheral transmission failure is a late-stage phenomenon in a long, moderate ride.

This isn't simply your brain giving up — there's evidence it's an active, protective process. In 8 male cyclists, central motor drive appeared to be adjusted in real time so that peripheral (muscle-level) fatigue never exceeded an individual threshold (Amann & Dempsey, Journal of Physiology, 2008) — your nervous system may be pre-emptively rationing effort rather than simply failing. Heat adds its own, separate central component: passive heating to a core temperature of 38.5°C reduced voluntary activation, which researchers attribute to "a failure of descending voluntary drive to compensate for changed muscle properties" (Todd, Butler, Taylor & Gandevia, Journal of Physiology, 2005) — though that study used passive heating rather than exercise, in a small sample of 7, so treat the exercise transfer as suggestive, not proven. A small conference-abstract comparison found more peripheral fatigue in recreational riders than in elite riders across a 3-hour protocol (Demay & Kerhervé, Journal of Science and Cycling, 2024) — interesting, but a conference abstract with a tiny elite sample, so we're grading it limited rather than moderate.

You're burning more oxygen for the same watts, and nobody has fully explained why

A separate, well-documented phenomenon is a straightforward loss of efficiency. In 14 well-trained male cyclists riding 2 hours at 60% of maximal minute power, gross cycling efficiency fell from 18.4 ± 1.6% to 17.4 ± 1.4% (p < 0.01) — meaning the same power output cost more metabolic energy at the end of the ride than at the start. At the same time, oxygen uptake in the working muscle actually rose, which the researchers describe as "suggestive of progressive mitochondrial or contractile inefficiency" (Hopker, O'Grady & Pageaux, Scandinavian Journal of Medicine & Science in Sports, 2017). An older study corroborates the whole-body picture: over 2 hours at 65% of maximal aerobic power, oxygen uptake rose 9.6% and heart rate rose 12.7% (Lepers, Hausswirth, Maffiuletti, Brisswalter & van Hoecke, Medicine & Science in Sports & Exercise, 2000), alongside a roughly 21% fall in freely chosen cadence and an 11–15% drop in peak pedal torque in the same study — signals you can actually feel and see on a head unit without a lab.

The important honesty here: the efficiency loss itself is a moderately well-replicated finding. The explanation offered for it — progressive mitochondrial or contractile inefficiency — is explicitly described by the researchers themselves as suggestive, not demonstrated. Rising oxygen cost at a fixed power is a real, repeated observation; why it happens at the cellular level is still a hypothesis.

The hunt for a biological marker of durability has come up empty

If durability were driven by a single identifiable muscle-tissue property, you'd expect to find it correlated with markers researchers already know how to measure. That search has, so far, failed. In 13 well-trained cyclists and triathletes, durability of the moderate-to-heavy exercise-intensity transition "was not related to vastus lateralis carnosine concentration, citrate synthase activity, or complex I activity" (Hamilton et al., European Journal of Applied Physiology, 2024) — three plausible candidate markers of muscle buffering capacity and mitochondrial function, all coming back null. The authors themselves note they may have been underpowered to detect a real effect, so this is a null result in a small sample, not proof that no such marker exists. But it's worth knowing that the tidy version of this story — "durability comes down to X measurable thing in your muscle" — currently has no confirmed X.

Putting the mechanisms in one place, honestly graded

No single mechanism explains durability, and that's not us being cautious for the sake of it — it's the explicit conclusion the evidence supports. Carbohydrate feeding attenuating critical-power decline is strongly supported. Whole-muscle glycogen depletion being the cause of that decline is directly contradicted by the correlation data. Fibre-specific glycogen depletion reducing oxidative capacity is a moderate, plausible mechanism with wide confidence intervals. Central nervous system fatigue arriving before peripheral transmission failure is a moderate, replicated finding, in old, small, male-only samples. Efficiency loss and rising oxygen cost are moderately well replicated as a phenomenon; the mitochondrial explanation for it is a hypothesis, not a demonstrated fact. And the search for an intramuscular biological signature of durability has, so far, turned up nothing.

If you take one sentence from this section: the honest overall grade here is mixed — moderate support for the phenomena themselves, limited support for why any of them happen. Anyone offering you a single, tidy mechanistic story for why your watts fade late in a ride is offering more certainty than the current evidence has earned.

What to actually do with this

None of this changes the basic, practical advice: fuel adequately during long rides, because carbohydrate availability is the best-supported lever you have, even though it doesn't operate purely by preventing an empty tank. Beyond that, three things follow directly from the mechanisms above.

  • Don't expect perfect fuelling to eliminate late-ride fade. Even elite cyclists taking in 100 g/hour still lost meaningful power late in a 4-hour ride. Carbohydrate buys you watts; it doesn't buy you immunity.
  • Treat cadence and pedal-feel changes as an honest signal. A falling freely-chosen cadence and a loss of torque "feel" late in a long ride are consistent with the efficiency and neuromuscular changes described above — they're not just you getting sloppy, they're a measurable physiological shift.
  • Be sceptical of any single "this is why you fade" claim, including ones about specific supplements or muscle markers targeting a single mechanism. The intramuscular marker hunt has come up empty so far, and this remains one of the more mechanistically unresolved corners of exercise physiology.

Common mistakes

The most common mistake is treating late-ride power loss as purely a fuelling failure, then concluding that because you fuelled "properly" and still faded, something is wrong with you. Based on the evidence above, that's not a reasonable conclusion — even well-fuelled elite cyclists lose power late in long rides, because carbohydrate availability is one contributing factor among several, not the whole story.

A second mistake is assuming a rising heart rate or rising perceived effort at constant power late in a ride is just fatigue in some vague sense, rather than the specific, measurable phenomenon it is: gross efficiency genuinely falls and oxygen cost genuinely rises for the same wattage. That's not being unfit; it's a documented physiological shift that happens to well-trained cyclists too.

A third mistake, common in supplement marketing, is pointing at a single muscle biomarker and claiming it explains durability. The one study that tested several plausible candidates — carnosine, citrate synthase, complex I activity — found none of them related to durability. Be wary of any product claim that leans on one of these markers as the mechanism it's fixing.

How to apply this week

On your next ride longer than two hours, pay attention to two free signals alongside your power number: your freely chosen cadence in the final third of the ride, and how much conscious effort it takes to hold a cadence you'd normally find automatic. If cadence has drifted down noticeably and holding it back up feels effortful rather than natural, that's consistent with the neuromuscular and efficiency mechanisms described here — a real physiological signal, not a mood. Log it alongside your power data for a few long rides and you'll start to see whether your own fade shows up first as a cadence problem, a power problem, or both together, which is more useful than any single number in isolation.

References

Frequently asked questions

Is late-ride power loss really not about running out of glycogen?

It's more nuanced than that. Feeding carbohydrate does slow power loss, so carbohydrate availability clearly matters. But when researchers measured muscle glycogen directly and compared it with the change in the power-duration relationship, the two barely correlated (r = 0.19 and r = 0.07 in one study), so whole-muscle glycogen level isn't the direct cause the intuitive story suggests.

What is the best-supported mechanism for late-ride power loss?

There isn't one dominant mechanism with strong evidence behind it. The most plausible current account is fibre-level: the most oxidative muscle fibres deplete their local glycogen first and effectively drop out, but even that evidence has wide confidence intervals and is graded moderate at best.

Does central (brain) fatigue happen before muscle fatigue?

In one well-cited study of 5-hour cycling, yes — voluntary activation fell within the first few hours while changes in the M-wave, a marker of peripheral transmission, appeared only after 4 hours. That's a moderate-grade, replicated pattern, but from small, male-only samples.

Why does my heart rate or perceived effort rise even when my power stays the same late in a ride?

Gross efficiency genuinely falls during prolonged exercise — one study found it drop from 18.4% to 17.4% over 2 hours — while oxygen uptake in the working muscle rises for the same power output. It's a documented physiological shift, not just fatigue in a vague sense.

Is there a muscle biomarker that predicts durability?

Not one that's been found yet. A study testing carnosine concentration, citrate synthase activity and complex I activity in well-trained cyclists and triathletes found none of them related to durability, though the study may have been underpowered.

If fuelling doesn't fully explain power loss, should I still fuel aggressively on long rides?

Yes. Carbohydrate feeding is the best-supported single lever for slowing critical-power decline during long rides, even though it doesn't work purely by preventing an empty tank and doesn't eliminate fade entirely.

Get The Forward

Research-led triathlon guidance, straight to your inbox. No spam, no filler — unsubscribe any time.