Skip to content

Threshold, durability, and long rides

Last checked: 2026-10-02

Scope. This brief answers questions 7, 9, and 10 for a trained amateur riding about 7–12 h/week, with goals of raising VO₂max, sustainable power (FTP and neighbouring constructs), and general fitness. It covers what threshold, sweet spot, tempo, endurance, and VO₂ work each do for sustainable power; how FTP relates to critical power (CP) and maximal lactate steady state (MLSS); which common interval recipes have actually been tested; how power-duration ability changes after large amounts of prior work; whether durability is separately trainable; and how to design the long ride, including fueling. It does not prescribe a full week. Intensity distribution, periodization, recovery spacing, and strength programming are owned by the other topic briefs. Durability as a performance determinant is only summarised here; the physiology brief should link here rather than repeat the late-ride evidence.

Navigation. Reference contents · CONTEXT · bibliography

Reading order: Previous — VO₂ intervals · Next — fueling and energy availability.

How to read this brief. initial_prompt.md refers to a transcript that is not included in the file. Statements such as “intervals were done early while fresh” are prompts about that missing transcript, not verified quotations. No paper below is attributed to the transcript. Candidate papers were identified by Europe PMC searches and, after access was restored on the check date, by web search. Quantitative results are taken only from retrieved abstracts, open full text, or one publisher PDF (Voet et al., 2025). Search snippets were not used as data. If a full text was not retrieved, the capsule is marked abstract-only. Sample sizes, baselines, and effect sizes that were not in the retrieved text are not filled in.

Intensities do not adapt exclusive physiological systems. A 20-minute power improvement can come from a higher VO₂max, a higher sustainable fraction, better efficiency, or less deterioration after prior work. Naming a session “threshold” does not mean it only trains a threshold system, and naming it “VO₂” does not mean it leaves FTP untouched.

Question navigation

Durable conclusions

Classification: Strong evidence

Claim. Group-level FTP, CP, MLSS, and the respiratory compensation point (RCP) correlate, but the limits of agreement are too wide, and the constructs are defined differently enough, that a single number should not be treated as all of them for one rider.

Support. In 15 trained and well-trained cyclists (VO₂max 62.3 ± 6.4 mL·kg⁻¹·min⁻¹), 95% of 20-minute power biased MLSS by only 1.4%, but the 95% limits of agreement were ±9.2% (Borszcz et al., 2019; abstract). In 23 trained men, 95% of 20-minute power and 60-minute power were close to an individual anaerobic threshold on average (bias −1 to −5 W), yet limits of agreement spanned tens of watts, and time to exhaustion at the 95% estimate was 50.9 ± 15.7 min, not 60 min (Borszcz et al., 2018; abstract). Twelve competitive men showed CP and FTP limits of agreement of +30 to −36 W (Morgan et al., 2019; abstract). A meta-analysis of nine studies (104 participants) found CP correlated with MLSS (r = 0.77) but MLSS sat about 11% below CP; VT1 sat about 30% below and RCP about 6% above (Galán-Rioja et al., 2020; abstract). A 2024 methods review argues that strict determination can shrink the CP–MLSS gap to a few watts, which is a reason to measure carefully, not a reason to swap casual field estimates (Caen et al., 2024; abstract). In 49 amateur triathletes, FTP and RCP correlated (r = 0.916) without a sex interaction on the slope, and were still not interchangeable (Engelke et al., 2026; abstract).

Contradictory or qualifying. McGrath et al. (2019; abstract) found that 17 of 19 highly trained athletes completed 60 min at 95% of 20-minute power in a quasi-steady state on a Wattbike, and that 5% above that power destabilised lactate. Agreement depends on population, ergometer, and the threshold method. Jones et al. (2019; full text) argue that conventional MLSS underestimates the maximal metabolic steady state and that CP is the heavy–severe boundary, with exercise at CP often tolerable for only about 20–30 min. Caen et al. (2024) show that some of the published gap is methodological. Both papers warn against naive interchange; they disagree on which index is the gold standard.

Practical implication. Use one anchor consistently for training zones, and retest it the same way. Do not convert a 20-minute test to “CP” or “MLSS” for an individual by multiplying by 0.95 and then prescribing to the watt. If the goal is the highest steady oxidative rate, a formal CP or MLSS test is more interpretable than FTP. If the goal is a repeatable field performance, the 20-minute or 60-minute mean itself is the outcome, and the 95% rule is only a rough estimate.

Uncertainty. Most validity samples are small and male. Sex-specific agreement is barely tested (Engelke et al., 2026, is the clearest recent check, and it is triathletes, abstract-only). Day-to-day noise of a few watts is smaller than the between-method limits of agreement, so the disagreement is not just test noise.

DUR-C02 — Sustainable power is not an exclusive product of threshold intervals

Classification: Moderate evidence

Claim. Work near the heavy–severe boundary, severe-domain intervals, and a background of easier riding can all move 20–40 min power. A week that is heavy in moderate/threshold time is a viable stimulus, not a proven best stimulus, and not a separate biological system from VO₂ work.

Support. In recreational cyclists (~6 h/week, VO₂peak 52 ± 6 mL·kg⁻¹·min⁻¹), 7 weeks of 4×8 min at the highest tolerable intensity improved VO₂peak, power at VO₂peak, and power at 4 mmol·L⁻¹ more than 4×4 min, 4×16 min, or low-intensity only (Seiler et al., 2013; abstract). The 4×16 min sessions were the most “threshold-like” (88 ± 2% HRpeak, 4.9 mmol·L⁻¹) and were not the best. In elite cyclists (VO₂max 73 ± 4), 3 weeks of effort-matched 30/15 s intervals raised 20-minute power 4.7 ± 4.4%, while 4×5 min intervals changed it −1.4 ± 2.2% (Rønnestad et al., 2020; abstract). Highly trained cyclists and triathletes (VO₂peak 64.5 ± 5.2) improved 40 km time-trial time 4.4–5.8% after 4 weeks with two weekly high-intensity sessions at peak aerobic power or 175% of peak power, versus little change in controls who kept only usual easy riding (Laursen et al., 2002; abstract). A 12-week load-matched comparison of 4×16, 4×8, and 4×4 min in different mesocycle orders improved 40-minute power 5–10% in all groups, with no order effect (Sylta et al., 2016; abstract). A concentrated block of moderate intervals (six sessions in 7 days, 5–7 × 10–14 min at RPE 14–15, then 6 days easy) improved power at 4 mmol·L⁻¹ by 4.0 ± 4.4% versus −1.3 ± 3.7% in time-matched regular preparation in 30 well-trained men (VO₂max 70.5 ± 4.6; Mølmen et al., 2025; abstract).

Contradictory or qualifying. Neal et al. (2013; abstract), a randomised crossover in 12 trained male cyclists, found a polarised 6-week block (about 80/0/20, 6.4 h/week) improved peak power, lactate threshold, and high-intensity capacity more than a threshold model (about 57/43/0, 7.5 h/week). Both models improved 40 km time-trial performance. Rosenblat et al. (2019; abstract) pooled time-trial results from three randomised trials and found a moderate effect favouring polarised over threshold (ES −0.66), with low study quality. Cove et al. (2025; abstract only in this brief) pooled 41 cyclist studies (797 participants, VO₂max ≥ 59 mL·kg⁻¹·min⁻¹) and found no significant difference between groups the authors labelled polarised and non-polarised for VO₂max or time-trial performance. Longer interventions were associated with larger VO₂max gains; extra volume beyond a necessary amount was not clearly additive. That abstract result is not a polarised-versus-pyramidal or 80/20 comparison. The distribution brief's full-text reading, not re-extracted here, found that Cove classified polarised versus non-polarised by whether intervals were shorter than 5 minutes, not by measured 80/20 or pyramidal time in zone. A separate 2025 individual-participant-data network meta-analysis compares training-intensity distributions using heart-rate time in zones and is the better polarised-versus-pyramidal evidence; methods are in the intensity-distribution brief, and this brief did not retrieve that PDF. Helgerud et al. (2007; abstract) found work-matched 4×4 min running raised VO₂max more than continuous lactate-threshold running in moderately trained men. That is a VO₂max result in runners, not a cycling FTP null result.

Practical implication. At 7–12 h/week, do not build the week as “threshold for FTP, VO₂ for VO₂max, endurance for endurance” as if the adaptations cannot overlap. Keep one or two high-quality hard sessions, some of which can be severe-domain work that also moves 20-minute power, and keep the rest easy enough that those sessions are actually hard. A moderate-interval block can raise power at 4 mmol·L⁻¹ in already well-trained cyclists, but the retrieved trial used a 7-day shock plus recovery, not a standing diet of sweet-spot hours.

Uncertainty. The cyclist trials are small, mostly male, and short (3–12 weeks). Cove et al. (2025) does not show that a measured polarised distribution is unnecessary, because its polarised/non-polarised split was not a measured 80/20 or pyramidal classification. It also does not identify an optimal threshold recipe. For polarised versus pyramidal, use the intensity-distribution brief's 2025 heart-rate time-in-zone network meta-analysis rather than this abstract.

DUR-C03 — The usual FTP menu, including sweet spot’s fatigue advantage, is coaching practice

Classification: Mostly coaching practice

Claim. 2×20, 3×15, 3×20, 4×10, 4×15, continuous 30–60 min, over-unders, and a uniquely efficient “sweet spot” band have not each been established by a controlled trial. The percentages coaches attach to sweet spot are definitions, not measured optima.

Support. Europe PMC searches on 2026-10-02 for sweet spot paired with FTP, and for the named recipes including over-unders, returned no relevant cycling training trials. A restored web search the same day likewise surfaced coaching explanations and the polarised-versus-threshold trials, not a randomised comparison of those recipes. The tested formats that are nearest the menu are different sessions: Seiler’s 4×4, 4×8, and 4×16 min; Stepto’s 12×30 s through 4×8 min; Rønnestad’s 30/15 s versus 4×5 min; Sylta’s 4×4 / 4×8 / 4×16 organisation; Mølmen’s 10–14 min moderate repeats in a block; Helgerud’s continuous lactate-threshold run. None is 2×20 at FTP versus 3×15, and none is over-unders versus steady threshold. See the recipe table below.

Contradictory or qualifying. Absence of a trial is not evidence that the sessions do nothing. Athletes clearly improve 20- and 40-minute power when they accumulate hard aerobic work in several formats (Sylta et al., 2016; Laursen et al., 2002; Seiler et al., 2013). The unsupported claim is the fine ranking: that one recipe, or the sweet-spot band, has a uniquely favourable stimulus-to-fatigue ratio.

Practical implication. Choose a format the athlete can complete at a high, repeatable quality, progress the accumulated time or the proximity to CP/MLSS only while the repetitions stay even, and do not change 2×20 to 3×15 because a platform ranks one as more efficient. Denham et al. (2026; abstract) found that 30 min at 85% of FTP already produced vigorous physiological and subjective strain in 12 healthy men, and that FTP alone underestimated demand at or above 85%. A “just below FTP” band is not automatically low fatigue.

Uncertainty. Coaching ranges such as 88–94% of FTP appear in secondary descriptions, not in a trial arm retrieved for this brief. They are not used here as physiological boundaries.

DUR-C04 — Power at the intensity boundaries falls after prior work, and 1500 / 2000 / 3000 kJ are not universal cut-points

Classification: Strong evidence

Claim. Critical power, power at the moderate-to-heavy transition, and short time-trial power decline after prolonged or high-intensity prior cycling. The same absolute kilojoule total does not mean the same fatigue. Intensity of the prior work, and work relative to body mass, change the drop.

Support. After 2 h of heavy cycling, 3-minute end-test power was unchanged at 40 and 80 min and about 9% lower at 2 h (260 ± 37 W to 236 ± 47 W) in 16 participants; 60 g·h⁻¹ carbohydrate preserved that estimate but not work above it (Clark et al., 2019, J Appl Physiol; abstract). A separate biopsy study in 14 participants found CP/end-test power about 11% lower and W′ about 20% lower after 2 h, with no meaningful correlation to the glycogen change (Clark et al., 2019, Am J Physiol; abstract). Jones (2024; abstract) summarises individual CP changes after 2 h heavy cycling from under 1% to about 32% around a 10% group mean. In athletes close to the 7–12 h range, 2 h at 90% of VT1 power dropped VT1 from 217 ± 42 to 196 ± 42 W and raised the heart rate at that transition (Stevenson et al., 2022; abstract; 14 cyclists/triathletes, 13 men, 1 woman, VO₂peak 59.9 ± 6.8). One hundred and fifty minutes of moderate cycling dropped VT1 power 6 ± 7% and 5-minute power 9 ± 10% in 13 athletes training 12 ± 2.1 h/week (Hamilton et al., 2024; abstract). After 1000 kJ at 80% of fresh 20-minute power, 20 amateur men (8.2 ± 2 h/week) lost 10.1 ± 6.5% on 20 min and 10.8 ± 7.8% on 5 min (Barsumyan et al., 2025b; full text). In 17 male professionals, work above CP of only 2.5, 5.0, and 7.5 kJ·kg⁻¹ lowered CP by 2.2%, 6.1%, and 16.2%; work-matched efforts below CP did not produce the same impairment (Mateo-March et al., 2024; abstract). A 2025/2026 systematic review of 21 acute studies concluded that prior work above CP produced roughly 10–20% power drops after 2.5–15 kJ·kg⁻¹, whereas below-CP work often produced smaller drops even at higher volumes, so kilojoules alone are an insufficient fatigue metric (Sánchez-Jiménez et al., 2025; full text; literature search 18 September 2024).

Contradictory or qualifying. The percentage drop is protocol-specific. Barsumyan’s 1000 kJ preload (about 13–14 kJ·kg⁻¹ in their ~74 kg sample) at a sub-FTP intensity produced ~10% losses in amateurs. Mateo-March’s professionals lost little CP until work was accumulated above CP, and then lost a lot at a small mass-specific dose. Both can be true. Field record-power profiles also mix racing, fueling, terrain, and drafting (Mateo-March et al., 2022; abstract). Professional durability estimates from the two best efforts in a season were reasonably repeatable, especially for longer efforts, and less so for 10 s (Mateo-March et al., 2025; abstract). That is reliability of a field metric, not proof of a training method.

Practical implication. For a 70 kg rider, 1500 kJ is about 21 kJ·kg⁻¹, 2000 kJ about 29 kJ·kg⁻¹, and 3000 kJ about 43 kJ·kg⁻¹. A lighter rider hits the same mass-specific load at fewer absolute kilojoules. Do not copy a professional “after 3000 kJ” benchmark as the amateur test or the amateur workout target. Report prior work as kJ·kg⁻¹ and describe whether it was below or above CP/FTP. A 2–2.5 h moderate ride is already long enough to move the first threshold and 5-minute power in athletes training about 12 h/week (Stevenson et al., 2022; Hamilton et al., 2024).

Uncertainty. Many laboratory samples are small. Female cyclists are scarce in the field databases. Heat, fueling, and the stochastic pattern of the preload are inconsistently controlled (Hunter et al., 2025; Peeters et al., 2025; both full text).

DUR-C05 — Fresh FTP or VO₂max does not fully identify who still has power late

Classification: Moderate evidence

Claim. Riders with similar fresh profiles can diverge after prior work. That does not mean durability is statistically independent of all aerobic traits in every population, and it does not mean a late-race result can be predicted from the fresh test.

Support. Barsumyan et al. (2025b; full text) found no significant regression of the 5- or 20-minute drop on VO₂max, relative FTP, training volume, years in sport, or age in 20 amateur men; the models were underpowered (adjusted R² −0.11 and 0.23). Ørtenblad et al. (2024; abstract) found a 10% drop in 6-minute power (95% CI 4 to 15) and a 6% drop in peak power after 4 h intermittent cycling with 100 g·h⁻¹ carbohydrate in 12 elite cyclists (VO₂peak 73.6 mL·kg⁻¹·min⁻¹), unrelated to fresh LT1, gross efficiency, maximal fat oxidation, or VO₂peak, and unrelated to fat oxidation during the protocol. Hamilton et al. (2024) found the 5-minute drop correlated with the VT1 drop (rs = 0.719) but not with resting vastus lateralis carnosine, citrate synthase, or complex I activity. Gallo et al. (2022; abstract) found that high-ranked juniors were separated by fresh record power, whereas high-ranked under-23 and professional climbers were separated by a smaller drop after 20–50 kJ·kg⁻¹. Successful versus less successful amateurs in Barsumyan et al. (2025a; full text) differed in the 20-minute drop after 1000 kJ (adjusted means 6.5% versus 12.5%) but also differed in body mass, VO₂max, and weekly hours; the 5-minute drop did not differ. The study was smaller than its own power calculation.

Contradictory or qualifying. Spragg et al. (2023a; abstract) reported a strong correlation (r = 0.891) between the fatigued-to-fresh CP ratio and relative peak power in 10 very young professionals (age 19.2 ± 0.8 y, VO₂max 74.4 ± 4.8, CP 5.6 ± 0.6 W·kg⁻¹) after 5×8 min at 105–110% CP. The abstract was truncated before the full correlation list and before the watt drop. Fresh aerobic traits sometimes track durability and sometimes do not. Protocol and population differ. Jones and Kirby (2025; full text) treat resilience as an added dimension precisely because start-line VO₂max, economy, and thresholds deteriorate differently across athletes. They also say the mechanistic basis is unresolved.

Practical implication. A higher fresh FTP is still worth having. It is the numerator that late-race power is a fraction of. It is not a substitute for occasionally measuring a hard effort after a relevant preload if the target event is decided after several hours. Do not tell an amateur that a disappointing fresh test “doesn’t matter because durability is independent.” In the amateur data, the successful group was also the group with higher VO₂max and more weekly training (Barsumyan et al., 2025a).

Uncertainty. Independence claims from n = 10–20 regressions are fragile. No retrieved study partials out years of training, muscle damage, heat, and carbohydrate availability at once.

DUR-C06 — Durability can improve; the best endurance session, and an effect independent of fresh fitness, are not established

Classification: Plausible but uncertain for the best endurance session and for an effect independent of fresh fitness. The strength addition itself is moderate evidence for that one fatigued outcome.

Claim. Durability can improve. Adding heavy strength twice weekly for 12 weeks improved 5-minute power after 185 minutes of cycling in one controlled study of well-trained cyclists (Rønnestad et al., 2011). That is causal evidence for a training addition, not proof the gain was independent of fresh fitness. The optimal endurance-session type remains uncertain. No matched trial of fresh versus late intervals was retrieved.

Support. The strength addition above is the controlled cyclist result (moderate evidence for that outcome). The clearest prospective endurance-training signal is observational, not a comparison of session types. Ten male semiprofessional riders (about 16.7 h/week, ~81% of time below LT1) improved both fresh and fatigued 1- and 10-minute power over 8 coach-led preseason weeks. Fatigued 10-minute power rose 25 W (95% CI 18 to 33; 7.3%), fresh 10-minute power 21 W (8 to 34; 6.4%). No single training-load metric predicted who improved. The authors explicitly say the data do not show that a polarised high-volume pattern had a direct effect on fatigued time-trial power (Voet et al., 2025; full text). A related season series in 16 semiprofessional men found 1-minute power more sensitive to a ~38 kJ·kg⁻¹ preload than 10-minute power, and described durability as improving more than fresh maximal power across the season; the retrieved abstract does not contain the percentage changes (Voet et al., 2024; abstract). In untrained and recreationally active adults, 10 weeks of either ~6.8 h/week low-intensity cycling or ~1.6 h/week high-intensity cycling improved a composite durability score during 3 h at 48% of pre-training VO₂max, with no significant group interaction (Matomäki et al., 2023; full text). Drifts did not change by a large amount. That population is not a trained amateur. Adding heavy strength twice weekly for 12 weeks improved 5-minute power after 185 min at 44% of maximal aerobic power from 371 ± 9 to 400 ± 13 W in well-trained cyclists, with no change in the endurance-only group, and lowered oxygen cost, heart rate, lactate, and RPE in the last hour (Rønnestad et al., 2011; abstract). That is a strength result, not an interval-placement result. A closely related report found higher 40-minute power and power at 2 mmol·L⁻¹ without a body-mass increase (Rønnestad et al., 2010; abstract). These may be the same cohort reported twice; they are not two independent replications.

Contradictory or qualifying. Spragg et al. (2023b; abstract) found, in 30 under-23 professionals, that time below VT1 and a shift toward a polarised distribution correlated modestly with a better fatigued 12-minute power (r about 0.38–0.43). Absolute fatigued 5- and 12-minute power were lower late season than early or mid season. Observational associations changed direction across the season. Jones and Kirby (2025; full text) state that, to their knowledge, no longitudinal study has identified the best training programme for resilience. They infer from elite practice and from cross-sectional comparisons, and they ask for trials that measure performance after a preload. Barsumyan et al. (2025a; full text) write that specific trainability remains an open question even while suggesting late race-pace efforts as practice.

Practical implication. Keep training, including enough easy volume to support quality and enough hard work to raise the fresh ceiling. Expect fatigued power to move, and do not expect a named “durability session” to be the proven lever. If strength is added, judge it by whether the endurance sessions stay high quality; the 2011 trial is not a licence to add late threshold on top of new lifting in a 7–12 h week. Strength programming itself belongs in the strength brief.

Uncertainty. Strength improved a fatigued 5-minute effort; it does not identify the best endurance session, and a related report from the same group also found higher fresh 40-minute power (Rønnestad et al., 2010), so independence from fresh fitness is unresolved. No matched fresh-versus-late interval trial was retrieved. Matomäki et al. (2023) show trainability of drift in untrained people and must not be transferred as “6.8 easy hours beats intervals” or the reverse. Voet’s riders already trained ~17 h/week under coaches. Their improvement does not identify which ingredient worked.

DUR-C07 — Late-ride intervals are unproven against the same intervals done fresh, and may raise recovery cost

Classification: Mostly coaching practice

Claim. There is no controlled evidence that “fatigue-state intervals” improve late-race performance more than fresh intervals plus separate long rides. Doing the hard work late makes the stimulus more event-like and can increase recovery cost. That higher cost is a plausible risk, not a result from a matched early-versus-late recovery trial.

Support. Every acute study above shows that the same nominal interval, done after prior work, is performed at lower power or that the boundary the athlete is targeting has already moved down. A late 5-minute effort in Hamilton’s athletes was 302 ± 63 W rather than 333 ± 75 W. A late 20-minute effort in Barsumyan’s amateurs was about 10% down. The session is therefore not the same physiological dose as the fresh version. Jones and Kirby (2025) found no trial of the best way to train resilience. Peeters et al. (2025; full text) review race-simulation protocols and conclude that mechanisms and improving interventions are not established; their search (7 January 2024) was about how to test, not about a winning workout. The 2026 systematic review’s suggestion that 5×8 min at 105% CP could build Grand Tour resilience cites Spragg’s fatigue test, not a training outcome (Sánchez-Jiménez et al., 2025). Using a test protocol as a weekly session is an extrapolation.

Contradictory or qualifying. Event-specific practice can still be useful for pacing, fueling, and decision-making even if it is not the best fitness stimulus. Nothing retrieved says late efforts are harmful in principle. The cost is opportunity. At 7–12 h/week there are usually only one or two high-quality slots. Filling one of them with a long ride plus a degraded interval can displace the fresh session that has clearer evidence for raising VO₂max and 20-minute power.

Practical implication. While fresh FTP and VO₂max are still the limiter, put the hard session first in the ride or on its own day. Introduce a late insert only after the athlete can finish the target long-ride duration fueled, below the first threshold, and still train well within 48 h. Treat the late insert as a replacement for a hard session, not an addition, and stop the progression if the next quality day collapses. Do not start a 7 h/week athlete with 5×8 min above CP at the end of a long ride.

Uncertainty. A matched-load trial of early versus late placement, with a fatigued time trial as the outcome, does not appear in the retrieved literature. Until it does, both “always fresh” and “always late” are over-precise.

DUR-C08 — There is no proven optimal long-ride duration or weekly share

Classification: Mostly coaching practice

Claim. No retrieved trial identifies an optimal long-ride duration, shows that extending one ride beats splitting the same weekly hours, or establishes what fraction of a 7–12 h week the longest ride should be. A share of the week is a scheduling budget so that quality and recovery still fit, not an optimum.

Support. Web search on 2026-10-02 found no 2024–2026 randomised cyclist trial of one continuous long ride versus the same endurance volume split, for chronic performance. The closest controlled comparison is an acute running crossover: 14 well-trained men (VO₂max 69.2 ± 4.2) did 6×10 min moderate once, or the same work as two 3×10 min sessions 6.5 h apart. The single session drifted more (heart rate 168 ± 7 to 173 ± 7 bpm; lactate 2.60 ± 0.75 to 3.01 ± 0.81 mmol·L⁻¹) and cost more next-morning soreness (Kjøsen Talsnes et al., 2024; abstract). The authors call this a starting point for “double threshold” organisation, not a cycling adaptation trial, and they note that continuous duration is more specific to road racing than to many running events. Zanini et al. (2025; abstract) showed that 90 and 120 min of heavy-domain running in 14 trained marathoners (VO₂max 63.1 ± 5.8; marathon about 2:47) lowered VO₂peak 3.1% and 7.1% and worsened oxygen cost 4.2% and 5.8%. That is an acute running time course, not a training prescription. Sitko et al. (2025; abstract), an expert panel rather than a trial, place “zone 2” immediately below the first lactate or ventilatory threshold and allow continuous, variable, or interval formats. They also say the expected adaptations may not be unique to that exact intensity.

Contradictory or qualifying. Specificity is plausible. Athletes who regularly do long runs have been described as preserving economy better, but the quantitative comparison Jones and Kirby (2025) cite includes an unpublished thesis and was not re-extracted here. Gallo et al. (2022) and the professional field studies show that people who race and train for many hours decline less after large mass-specific loads. That is not a volume-matched trial of 4 h versus 2×2 h.

Practical implication. Budget the longest ride so it does not consume the week. Any percentage is a coaching budget, not a result, and 35% is not a validated ceiling. A broad untested span is about 25–40% of cycling time. One worked example, which already keeps strength inside the 7/9/12 h total rather than adding it on top, is a long ride of 120 min inside 390 cycling minutes at 7 h (~31%), 180 min inside 480 cycling minutes at 9 h (~38%), and 210 min inside 660 cycling minutes at 12 h (~32%). Other durations inside that span are equally unproven. Extending that ride by deleting another easy ride (same weekly total) is not known to be better than keeping two medium rides. Extending it by adding hours is a volume question, owned by the volume brief, and it has a recovery cost the acute running comparison makes visible. Variable terrain is acceptable if power stays below the first threshold. Repeated climbs that push above it are unplanned threshold, not a better zone 2.

Uncertainty. The budget will be wrong for some athletes and some events. A first long road race may justify one longer ride in the final weeks as specific practice, even if that week is a poor general-fitness week. That is event preparation, not the year-round optimum.

DUR-C09 — Carbohydrate availability modulates durability; it is not structural durability, and fasted long rides are not the method

Classification: Moderate evidence

Claim. Low carbohydrate availability can lower power at the moderate-to-heavy transition and gross efficiency, and carbohydrate ingested during prolonged moderate cycling can attenuate the fall in estimated CP. The same literature shows a remaining drop, a fall in W′ that carbohydrate does not fix, and performance losses despite high in-ride carbohydrate. Preserving glycogen or raising fat oxidation is not the same adaptation as resisting the fall in power.

Support. Nine well-trained women, after glycogen-depleting exercise and either ≥9 or ≤1 g·kg⁻¹ carbohydrate over the next day, had lower VT1 power in the low condition (133 ± 24 versus 152 ± 28 W) and lower gross efficiency, with no difference in 3-minute end-test power (Main et al., 2026; abstract). Sixteen endurance-trained cyclists and triathletes (VO₂max 51.8 ± 6.8) rode 3 h at 95% of the estimated gas-exchange threshold (GET), the first threshold and a moderate-domain intensity, not at FTP, after a standardised high-carbohydrate day (8 g·kg⁻¹ plus a 2 g·kg⁻¹ pre-exercise meal). Estimated CP was 277 ± 27 W fresh, 236 ± 30 W with water, 257 ± 28 W with 60 g·h⁻¹, and 266 ± 29 W with 120 g·h⁻¹ (1:0.8 maltodextrin:fructose). W′ fell without a carbohydrate-condition difference (Norte et al., 2026; full text). Clark et al. (2019) had already shown that 60 g·h⁻¹ preserved the CP estimate after 2 h heavy work without preventing the glycogen decline, and that the glycogen change did not correlate with the CP or W′ change. Ørtenblad et al. (2024) still observed a 10% 6-minute loss after 4 h with 100 g·h⁻¹. Aird et al. (2018; abstract) found pre-exercise feeding helped prolonged aerobic performance and that fasted exercise raised post-exercise free fatty acids; chronic performance benefit of fasting was not established. Marquet et al. (2016; abstract) periodised carbohydrate timing at the same daily intake (6 g·kg⁻¹) in 21 triathletes for 3 weeks. Sleep-low improved submaximal cycling efficiency more than control (+11 ± 15% versus +1.4 ± 9.3%) and improved some performance tests. Gejl et al. (2017; abstract) found no superior effect of 4 weeks of periodic carbohydrate restriction in 26 elite men; both groups improved VO₂max and 30-minute power by about 5%.

Contradictory or qualifying. Sleep-low can help selected outcomes when daily carbohydrate is adequate and the hard session is fueled (Marquet et al., 2016). It did not add anything detectable in elites already training hard (Gejl et al., 2017). Neither study measured durability as Maunder defined it (shift in profiling variables during prolonged exercise). Impey et al. (2018; abstract) offer a “fuel for the work required” framework and a glycogen-threshold hypothesis. That is a mechanistic proposal, not a performance trial in trained cyclists. Hunter et al. (2025) and Peeters et al. (2025) both warn that nutrition before and during the preload is a confounder in durability tests, which is another reason not to treat a fasted laboratory drop as the athlete’s structural durability.

Practical implication. Fuel the long ride and the quality work. For a rider who already tolerates carbohydrate, something in the range tested during prolonged cycling (60 g·h⁻¹, and in one small 2026 trial up to 120 g·h⁻¹) reduces the CP drop without abolishing it. Do not use a fasted or severely restricted long ride as the durability programme. Do not confuse one delayed breakfast with chronic low energy availability. The latter is a health problem, not a zone. Gut symptoms were an outcome in Norte et al. (2026), not a blank. In that sample, 120 g·h⁻¹ raised cumulative symptom score and peak regurgitation, fullness and cramps versus water, while 120 versus 60 g·h⁻¹ was not significant, and 90 g·h⁻¹ was not tested. Do not treat 120 g·h⁻¹ as universally tolerated or universally intolerable. Progress intake if it is new. This brief does not set a medical carbohydrate prescription.

Uncertainty. Norte’s sample had a VO₂max of about 52 mL·kg⁻¹·min⁻¹, useful for trained amateurs and not a WorldTour fueling trial. The preload was 95% of GET, not FTP. Main et al. (2026) is the main female cyclist crossover retrieved; n = 9, and it is severe next-day carbohydrate restriction after depletion, not a skipped breakfast. Norte did record gut symptoms (section 3.5, Figure 7): six symptoms every 30 min. Only 4 of 16 participants scored ≥4 and below 8, in the 60 or 120 g·h⁻¹ conditions. Cumulative scores were higher at 120 g·h⁻¹ than water (p = 0.005; medians 18 [0–81] versus 6 [0–56]) but 120 versus 60 was not significant, nor was 60 versus water. Peak regurgitation, fullness and cramps were higher at 120 than water, again with no 120-versus-60 difference. Ninety grams per hour was not tested. That is not universal intolerance, and it is not evidence that every rider tolerates 120 g·h⁻¹.

Q7. Threshold, FTP, and sweet spot

What the anchors are

FTP, in the original practical definition, is the highest power a rider can hold in a quasi-steady state for about an hour (Borszcz et al., 2018). The common field shortcut is 95% of a 20-minute time trial. CP is the asymptote of the power–duration relationship, theoretically the highest power at which a metabolic steady state can still be achieved; W′ is the finite work that can be done above it (Poole et al., 2016; abstract; Jones et al., 2019). MLSS is the highest power at which blood lactate can be stabilised under a specified criterion, usually a series of 30-minute constant-load trials. RCP and the second ventilatory threshold are gas-exchange estimates in the same neighbourhood, not the same number (Galán-Rioja et al., 2020).

Anchor What a careful version estimates What the common field shortcut does Individual error retrieved
FTP 95% of 20 min Roughly hour power / near MLSS on average One hard 20-min effort, times 0.95 LoA about ±9% vs MLSS; TTE at that power ~51 ± 16 min in one sample; 17/19 completed 60 min in another
60-min mean power The original FTP definition Expensive to repeat MacInnis et al. (2019): 20-min tests were reliable (typical error 4.6 W) in 8 trained men; 60-min mean was 309 ± 26 W against 20-min means near 343 W, so 95% of 20 min sat above measured hour power in that small sample
CP Heavy–severe boundary if the trials are valid Often fitted from too few or poorly spaced efforts LoA vs FTP +30 to −36 W (Morgan et al., 2019). Caen et al. (2024): gap vs MLSS can shrink to a few watts with strict methods
MLSS Highest lactate-stable power under the study’s rule Sometimes inferred from one incremental test About 11% below CP across the 2020 meta-analysis; not a reason to ignore it, a reason not to rename it CP
RCP Gas-exchange point often above CP Read off one ramp About 6% above CP in the meta-analysis; strongly associated with FTP in amateur triathletes, not interchangeable (Engelke et al., 2026)

MacInnis et al. (2019) is abstract-only, n = 8, one 60-minute trial. The comparison of 95% of the 20-minute means with the 60-minute mean is arithmetic from the reported means, not a Bland–Altman test the authors presented. It is enough to reject a universal 95% rule, not enough to publish a new correction factor.

Roles, without exclusive systems

Severe-domain / VO₂-type work raises the ceiling and, in cyclists, often the 20- and 40-minute power that athletes call FTP. The formats that did this in retrieved trials were short (30/15 s; efforts at peak aerobic power), not traditional threshold reps (Rønnestad et al., 2015, 2020; Laursen et al., 2002; Stepto et al., 1999). Seiler et al. (2013) found the middle duration, 4×8 min, better than both 4×4 and 4×16 for VO₂peak and power at 4 mmol·L⁻¹ in recreational cyclists. Duration and intensity interacted. More minutes near threshold was not automatically better.

Threshold / upper heavy domain is time at a high fraction of VO₂max with less non-steady-state disturbance than work above CP. It can raise lactate-threshold power and time-trial performance (Neal et al., 2013; Mølmen et al., 2025; Sylta et al., 2016). It is not uniquely required, and a distribution that puts ~43% of training time in the moderate zone was outperformed by a polarised distribution in one 6-week crossover (Neal et al., 2013) Cove et al. (2025) did not clearly separate groups labelled polarised and non-polarised, but that label was not a measured pyramidal or 80/20 distribution (see DUR-C02). It is not evidence that a threshold-heavy week matches a true pyramidal or polarised week.

Tempo and sweet spot sit between the first threshold and FTP in coaching models. Sitko et al. (2025) expect broad aerobic adaptations from work just below the first threshold and say similar adaptations may occur slightly above or below that line. That expert view cuts against both “tempo is a different system” and “only an exact zone 2 wattage works.” It does not validate sweet spot as the efficiency peak.

Long endurance supplies volume, technical time, and the only practical way to accumulate the hours that separate durable professionals from juniors in cross-sectional data (Gallo et al., 2022). Helgerud et al. (2007) should not be read as “easy riding does not adapt.” It was work-matched, short, and in moderately trained runners, and the outcome was VO₂max. At 7–12 h/week the constraint is that easy volume is scarce. Spending it as accidental tempo reduces the quality of the hard day. That is a programming interaction, not a claim that 55% of FTP is biologically inert. Denham et al. (2026) classified 55% of FTP as moderate and 85% as vigorous in healthy men over 30 min. Population: not described as trained cyclists.

Recipe comparison

Coaching recipe Retrieved as its own RCT arm? Closest tested analogue What that analogue showed Do not pretend
2×20 min near FTP No Sylta et al. (2016): 4×16 min maximal session effort, lactate 4.7 mmol·L⁻¹, inside a 12-week mixed programme, cyclists VO₂peak 61 ± 6 All organisations improved 40-min power 5–10%; order of 4, 8, and 16 min did not matter That 2×20 was compared with 3×15 or 4×10
3×15 min No None closer than the 10–16 min repeats above — A ranking versus 2×20
3×20 min No Continuous-ish accumulation of ~45–60 min near threshold is the idea, not a retrieved arm — That more continuous time is proven better
4×10 min No Stepto et al. (1999): 8×4 min at 85% peak aerobic power, ~4 riders per cell, 6 sessions in 3 weeks plus usual base Modelled 40 km gain about 2.8% (95% CI spanning roughly 1.3–4.3 points). 4×8 min at 80% peak power did not significantly improve That 4×10 at FTP was the winning arm. The significant formats were harder and shorter, or much shorter and supramaximal (12×30 s at 175% peak power, modelled +2.4%). n per cell is too small for a recipe ranking
4×15 min No Seiler et al. (2013) 4×16 min Improved less than 4×8 min for VO₂peak and power at 4 mmol·L⁻¹ in recreational cyclists That longer threshold reps are the FTP specialist format
Continuous 30–60 min No training RCT retrieved Helgerud et al. (2007): continuous running at lactate threshold (85% HRmax), work-matched, 8 weeks, moderately trained men. McGrath et al. (2019): 60 min test at computed FTP Continuous threshold running was inferior to 4×4 for VO₂max. Holding FTP for an hour is a test result in highly trained athletes, not a training outcome That a weekly 40- or 60-min steady FTP ride is the evidence-based progression
Over-unders No Acute W′ reconstitution models (Chorley and Lamb, 2020; narrative review) and the finding that work above CP costs more durability per kJ than work below CP (Mateo-March et al., 2024) Oscillating around CP spends and may partly restore W′. That is an acute description That over-unders build more FTP or more durability than steady work near CP

Mølmen et al. (2025) is the closest well-trained cyclist trial of long moderate intervals: 5–7 repetitions of 10–14 min at RPE 14–15, six times in one week, then recovery. Power at 4 mmol·L⁻¹ rose about 4% relative to ordinary preparation. It is a block, not a standing 2×20 prescription, and the abstract does not report every secondary percentage. It belongs in the “moderate intervals can work” column, not in a winner’s column for 3×15.

Stepto et al. (1999) is easy to over-read. Twenty male cyclists (VO₂peak 4.8 ± 0.6 L·min⁻¹) were split across five interval types. The cubic model favoured 85% and 175% of peak aerobic power. Intervals at 80% and 100% of peak power did not significantly improve the 40 km time trial. Peak aerobic power is not FTP. Eighty percent of peak aerobic power is often near or above MLSS, so the “threshold-like” arm in that study is the one that did not clearly work. With about four riders per arm, a null is not proof of uselessness. It is proof that this recipe was not established.

What seems to matter more than the recipe name

Proximity to the heavy–severe boundary changes the kind of disturbance. Below CP, a metabolic steady state is at least possible. Above CP, W′ is spent, lactate and VO₂ do not stabilise in the same way, and a small mass-specific dose already impairs subsequent power (Mateo-March et al., 2024; Poole et al., 2016). Prescription by %FTP inherits all the individual error in DUR-C01. A rider whose FTP sits 20 W above CP is doing severe-domain work in a “threshold” session. A rider whose FTP sits 20 W below CP is doing heavy-domain work. Heart rate, RPE, and whether the repetitions stay even are the practical checks when CP is unknown.

Accumulated time and total work matter, but not as “more threshold minutes always win.” Seiler et al. (2013) matched effort, not kilojoules. Thirty-two minutes at ~90% HRpeak beat 16 minutes at ~95% and also beat 64 minutes at ~88%. Intensity and duration interacted. Laursen et al. (2002) added two weekly high-intensity sessions on top of usual easy training in athletes who were already at VO₂peak 64.5; the control group that only continued easy training did not improve the 40 km trial. Some hard work was necessary. The study does not say how many threshold minutes were optimal.

Progression that is supported by the trials is coarse: add a hard session if the athlete has none (Laursen et al., 2002); keep session quality high enough that longer reps are not just slower junk (Seiler’s 4×16 min lost to 4×8); organise 4-, 8-, and 16-minute sessions in any order if the 12-week load is the same (Sylta et al., 2016); a short block of moderate intervals can move power at 4 mmol·L⁻¹ if it is followed by recovery (Mølmen et al., 2025). Progression by creeping every endurance ride toward 85–90% of FTP is the failure mode at 7–12 h, because it spends the recovery that the hard session needs and it was the pattern closest to Neal’s less effective threshold model.

RPE and even repetitions are the field version of “maximal tolerable intensity” used by Seiler and Sylta. Those studies did not prescribe a rigid %FTP. Power can be the record of the session. It is a poor master if the anchor is wrong.

Sweet spot

The efficiency claim is a model, not a measurement. Training-stress scores rise steeply with intensity, so a band just below FTP can accumulate a high score with less session failure than work at or above FTP. That arithmetic does not measure mitochondrial adaptation, time-trial gain, or next-day performance. No trial retrieved here isolated that band against matched work at FTP, against 4×8 min, or against polarised training with the moderate zone removed.

Denham et al. (2026) is the recent physiological check on two FTP-derived intensities, and it cuts the other way: 85% of FTP was already vigorous, and FTP-only zones underestimated strain. The participants were 12 healthy men, not a trained-cyclist cohort. Transfer is limited, but it is enough to refuse the slogan that sweet spot is a low-fatigue free lunch.

For a 7–12 h athlete, sweet spot is a way to accumulate hard aerobic time when a true severe session is not the priority that week. It is not the default endurance intensity, and it is not known to be more efficient per unit fatigue than a well-executed threshold or VO₂ session of similar stress.

Q9. Durability and training under fatigue

Measured declines

Maunder et al. (2021; abstract) defined durability as the time of onset and the magnitude of deterioration in physiological profiling characteristics during prolonged exercise. Jones (2024) argued that this deterioration is large enough, and individual enough, to sit beside VO₂max, economy, and fractional utilisation. Jones and Kirby (2025) prefer “physiological resilience” for the capacity and treat durability as the performance expression. Zanini, Jones, and Nybo (2025) published a definitional commentary; Europe PMC had no abstract on the check date, so its proposals are not used here.

Study Who Preload What fell Retrieved change Design / access
Clark et al., 2019, J Appl Physiol 16 participants Heavy cycling, 40 / 80 / 120 min; placebo or 60 g·h⁻¹ CHO 3-min end-test power (CP estimate) Unchanged at 40 and 80 min; −9% at 2 h (260 ± 37 to 236 ± 47 W). CHO: 254 ± 49 W, reduction negated. W′ 13.5 ± 3.4 kJ, still down Acute. Placebo time course from the Europe PMC abstract; CHO watts from the PubMed abstract read on 2026-10-02
Clark et al., 2019, Am J Physiol 14 participants 2 h heavy CP and W′ ~11% and ~20% lower; not correlated with muscle glycogen change (r = 0.19 and 0.07) Acute + biopsy; abstract
Stevenson et al., 2022 14 trained, 13 men, 1 woman, VO₂peak 60 2 h at 90% of VT1 power Power at moderate–heavy transition 217 ± 42 to 196 ± 42 W; heart rate at the transition rose (142 ± 9 to 151 ± 12 bpm) Acute; abstract. Open full text exists; numbers here are from the abstract
Hamilton et al., 2024 13 cyclists/triathletes, VO₂peak 57.3 ± 4.8, 12 ± 2.1 h/week 150 min moderate VT1 and 5-min TT VT1 211 ± 40 to 198 ± 39 W (−6 ± 7%); 5-min 333 ± 75 to 302 ± 63 W (−9 ± 10%) Acute; abstract
Barsumyan et al., 2025b 20 amateur men, 34.6 ± 7.3 y, 74.4 ± 8.7 kg, 8.2 ± 2 h/week 1000 kJ at 80% of fresh 20-min power, 60 g·h⁻¹ CHO 5- and 20-min power −10.8 ± 7.8% and −10.1 ± 6.5% Acute field; full text. Not a training study
Barsumyan et al., 2025a 14 amateurs, VO₂max 52.0 ± 7.4, 9.6 ± 2.2 h/week Same 1000 kJ idea 20-min drop by success group Adjusted 6.5% vs 12.5% (p = 0.033). 5-min: no group difference Cross-section; full text. Underpowered versus its own n = 26 plan
Ørtenblad et al., 2024 12 elite/professional, VO₂peak 73.6 4 h intermittent, 100 g·h⁻¹ CHO 6-min TT, peak power −10% [4, 15]; peak power −6% [0, 6] Acute; abstract
Mateo-March et al., 2024 17 male professionals, season-long observational 2.5 / 5.0 / 7.5 kJ·kg⁻¹ above vs below CP CP and MMP CP −2.2%, −6.1%, −16.2% after above-CP work. Below-CP matched work did not similarly impair Field observational; abstract
Mateo-March et al., 2022 112 male professionals, 2013–2021 0, 15, 25, 35, 45 kJ·kg⁻¹ Record MMP, 10 s to 120 min Decline with fatigue; category comparison stated as significant. Percent drops were not in the retrieved abstract Database; abstract-only for magnitudes
Norte et al., 2026 16 trained cyclists/triathletes, VO₂max 51.8 ± 6.8 3 h at 95% of estimated GET (first threshold, moderate domain, not FTP), after 8 g·kg⁻¹ CHO plus a 2 g·kg⁻¹ meal Estimated CP Fresh 277 ± 27 W; water 236 ± 30; 60 g·h⁻¹ 257 ± 28; 120 g·h⁻¹ 266 ± 29. W′ down in all, no CHO difference. 90 g·h⁻¹ not tested Acute crossover; full text
Zanini et al., 2025 14 male marathoners, VO₂max 63.1, ~2:47 90 and 120 min heavy-domain running VO₂peak, economy VO₂peak −3.1% and −7.1%; oxygen cost +4.2% and +5.8% Acute running; abstract. Not cycling
Sánchez-Jiménez et al., 2025 21 studies, mixed lab and field Various Power after prior work Above CP: about 10–20% drops after 2.5–15 kJ·kg⁻¹. Below CP: often <5% even at higher volumes, in the review’s synthesis Systematic review of acute effects; full text. Search closed 18 Sep 2024, so it does not include Norte 2026 or Voet 2025

The review’s summary of Leo et al. (2021) is second-hand here: under-23 riders showed mean-maximal-power losses in efforts of 12 min or less after 1000 kJ, with longer efforts declining around 1500–2500 kJ, whereas professionals’ non-5/12 min durations declined around 2000–3000 kJ. That sentence is the review’s, not a re-extraction of Leo. It is included because the question asks about those kilojoule bins, and because it illustrates the same point as Mateo-March et al. (2024): the bin moves with athlete level and with the intensity hidden inside the kilojoules.

Jones and Kirby (2025) describe the CP fall over 2 h as accelerating after about 80 min in the Clark time course, which matches “no change at 40 and 80 min, clear change at 2 h.” A 90-minute amateur ride is not automatically a durability session. A 2–3 h ride can be, even if it never reaches 3000 kJ.

Observational racer gaps are not a training method

Professionals and successful climbers decline less after large mass-specific loads than juniors or lower-ranked riders (Gallo et al., 2022; Mateo-March et al., 2022). Under-23 professionals’ fatigued power profile moved across a season more than the fresh profile, and time below VT1 plus a polarised shift correlated modestly with better fatigued 12-minute power, while late-season fatigued power was also worse in absolute terms (Spragg et al., 2023b). These are comparisons of people who already differ in years of racing, team support, and selection. They do not identify a workout.

The prospective cyclist evidence retrieved for this brief is:

  • Voet et al. (2025): coach-led training at ~17 h/week improved fresh and fatigued power; load metrics did not explain who improved; authors refuse a direct prescription.
  • Voet et al. (2024): across a season, durability described as improving more than fresh power; abstract lacks the percentages.
  • Rønnestad et al. (2011): heavy strength improved a 5-minute effort after 185 min. Not interval placement.
  • Matomäki et al. (2023): untrained adults improved drift with either easy volume or intervals. Do not export the hours to a trained amateur.

No arm trained late intervals against fresh intervals. No arm trained over-unders against steady threshold for a fatigued outcome. The 2026 review’s practical example of 5×8 min at 105% CP is a suggested mimic of a fatigue protocol (Sánchez-Jiménez et al., 2025). It is not evidence that the protocol builds durability, and it is a poor first session at 7–12 h/week.

Fresh ceiling versus late-race ability

Training to raise fresh FTP and VO₂max uses the hard work when power, cadence, and recruitment are highest. That is the stimulus in Seiler, Rønnestad, Laursen, and Sylta. Training to perform after several hours is a different goal: the athlete must still be able to operate near a falling CP, with falling efficiency (Stevenson et al., 2022; Main et al., 2026) and with W′ already reduced (Clark et al., 2019; Norte et al., 2026). Those are not solved by the same session on the same day unless the week has room.

What late intervals might produce, if they produce anything specific. Practice at pacing and fueling after prior work; a stimulus to whatever limits the individual (glycogen availability, fiber recruitment, neuromuscular tolerance, heat). None of those mechanisms has been tied to late-versus-early interval placement in a training trial. Hamilton et al. (2024) found no link between durability and three resting muscle markers. Ørtenblad et al. (2024) found no link to fat oxidation. Clark et al. (2019) found no link to the glycogen delta. The mechanism list in Jones and Kirby (2025) — fiber type and capillaries, glycogen, cardiovascular limitation, muscle damage, rising oxygen cost, psychology — is a hypothesis list. Strong mechanism language would overclaim it.

Recovery cost. A late insert stacks duration and intensity, so a higher recovery cost is a plausible risk, not a measured early-versus-late difference. No matched trial compared next-day performance after the same intervals placed early versus late. Talsnes et al. (2024) answers a different question: one long moderate running session versus the same work split had more drift and next-morning soreness. That does not prove a late cycling interval costs more recovery than the same interval done fresh. At 7–12 h, the next hard session is often 48 h away and is a large fraction of the week’s quality. If it is worse, the late insert has a negative transfer even if it felt race-specific.

When to introduce it. After the long ride itself is stable: the athlete finishes the budgeted duration fueled, mostly below the first threshold, without illness or a collapsed next quality day. After fresh hard sessions are already progressing. Before a target event that is actually decided after 2–4 h, as occasional specific practice, replacing a hard day. Not in the first block of a 7 h week whose limiter is still a low fresh 20-minute power.

Q10. Long-ride design

Duration, splitting, terrain, inserts

Optimal duration. Not established. Acute data say the moderate-to-heavy boundary and short-trial power are already down after 2–2.5 h of moderate cycling in trained amateurs (Stevenson et al., 2022; Hamilton et al., 2024), and CP is clearly down by 2–3 h of heavy or moderate work (Clark et al., 2019; Norte et al., 2026). A ride shorter than 90 min is a weak durability test. That does not make 90 min the minimum training dose, and it does not make 4–6 h optimal. Professional absolute durations are specific to events of 4–6 h and to weekly volumes this audience does not have.

Extending one ride versus splitting the same hours. No cyclist adaptation trial retrieved. Talsnes et al. (2024) suggests the continuous version is a larger acute internal load and a larger next-day cost, in runners, at matched moderate-interval time. For an amateur, splitting can be the way to keep intensity honest and protect the next day. One longer ride remains the specific practice for sitting on the bike, eating, and pacing for 3–4 h. Use the longer ride when the event needs it. Do not assume it grows a unique adaptation that two medium rides cannot, if weekly hours are fixed.

Terrain and intensity creep. Sitko et al. (2025) allow variable zone 2, provided the intensity stays immediately below the first threshold. Hills that force repeated surges above that line change the session into threshold. Stevenson’s heart-rate finding matters here: after 2 h, the heart rate attached to the first threshold was higher while the power was lower. Holding heart rate will cut power. Holding the starting power may push the rider above a falling threshold. Either pattern can look like “I was still in zone 2” on a file and be something else physiologically. RPE and a power cap relative to a fresh anchor, reviewed after the ride, are safer than a single metric. This is coaching translation of acute data, not a trial of flat versus rolling routes.

Where to insert work. Early, or separate, when the goal is the fresh ceiling. Late, occasionally, when the goal is event practice and the week can absorb a plausible extra recovery cost (DUR-C07), which has not been measured in a matched early-versus-late trial. A progressive rise in intensity across the long ride is a common elite pattern described by Jones and Kirby (2025) for some runners. It was not tested as a cycling intervention in the papers retrieved. If used, the early part has to be easy enough that the late part is the planned hard work, not a second unplanned threshold block.

Muscular fatigue. Durability loss is not only empty glycogen. W′ falls even when carbohydrate preserves CP (Clark et al., 2019; Norte et al., 2026). High carbohydrate intake did not stop a 10% 6-minute loss in elites (Ørtenblad et al., 2024). Resting oxidative-enzyme activity did not explain the VT1 drop (Hamilton et al., 2024). Cadence may fall more than torque in some fatigued record-power efforts; the 2026 review cites that pattern from other groups, and this brief did not re-extract those papers, so it is noted only as a hypothesis the review raises (Sánchez-Jiménez et al., 2025). Practically: a long ride that is also a neuromuscular smash (steep gears, long climbs above threshold, or a late VO₂ set) is a different stress from a steady aerobic ride. Count it as hard.

Fasted versus fueled, and the weekly share

Fuel the long ride if the goal is to complete it as aerobic volume or as a durability test that is not confounded by avoidable carbohydrate shortage (Hunter et al., 2025; Peeters et al., 2025; Norte et al., 2026). Aird et al. (2018) found that exercising after an omitted pre-exercise meal raised post-exercise free fatty acids and that feeding helped prolonged performance; that meta-analysis does not show a drop in cycling VT1. Main et al. (2026; abstract, PMID 42228164) is a different intervention: nine well-trained women did glycogen-depleting exercise, then ate either ≤1 or ≥9 g·kg⁻¹ carbohydrate over the next ~24 h before the tests. The low condition lowered VT1 power (133 ± 24 versus 152 ± 28 W) and gross efficiency. It does not isolate skipping breakfast. Sleep-low, with hard sessions still fueled and daily carbohydrate held at 6 g·kg⁻¹, improved some performance markers in trained triathletes over 3 weeks and did nothing extra in elites over 4 weeks (Marquet et al., 2016; Gejl et al., 2017). Neither is a durability-training trial. Chronic under-eating is a different intervention from session timing and is outside this brief’s prescription.

Recommended share for 7–12 h. There is no trial. The table is one scheduling budget so the longest ride does not erase the rest of the week, with strength already counted inside 7, 9, and 12 h. It is not an optimum and not a ceiling.

Total week, strength included Cycling time in this example Long ride in this example Share of cycling time Status
7 h 390 min 120 min ~31% One coaching example, not an optimum
9 h 480 min 180 min ~38% Same. Shows that ~35% is not a ceiling
12 h 660 min 210 min ~32% Same. Still far short of a 5–7 h professional ride

These three figures sit inside a broad coaching span of about 25–40% of cycling time. That span is not a validated band, and nothing retrieved makes 35% a maximum. Strength is already inside the 7/9/12 h total in this example; it is not extra time on top. The next day after the long ride should still be easy or rest if the ride was long enough to move the first threshold, and the week's hard session should not be stacked on the same day unless that is the planned quality. A ride that takes most of the cycling week remains a poor default because it crowds out quality, not because a trial forbade it.

Transfer

7–12 h versus 20–30 h

Professional durability is visible after 20–50 kJ·kg⁻¹ and after races of 4–6 h (Gallo et al., 2022; Mateo-March et al., 2022; the kilojoule bins in Sánchez-Jiménez et al., 2025). Those athletes also train enough hours that a 5–6 h ride can be 20–25% of the week, followed by more easy volume. Copying the absolute duration into a 10 h week makes the long ride half the week or more and usually forces intensity creep or a lost quality session.

The phenotype is already present at amateur volumes. Athletes at 8–12 h/week lose about 6–11% of 5- to 20-minute power after 1000 kJ or 2–2.5 h (Barsumyan et al., 2025a, 2025b; Hamilton et al., 2024; Stevenson et al., 2022). They do not need a WorldTour preload to have a durability problem. They also cannot absorb WorldTour solutions: 40 kJ·kg⁻¹ preloads, 5×8 min above CP at the end of a long ride, or 17 h weeks like Voet’s semiprofessionals.

What transfers: keep most riding below the first threshold so the hard day is hard (Seiler, 2010, is descriptive of successful endurance athletes, not a 7 h prescription; Sitko et al., 2025); use one or two quality sessions that can move 20-minute power even if they are not named threshold (DUR-C02); fuel long rides; judge progress by a repeatable hard effort, and occasionally by the same effort after a preload the athlete can actually recover from. What does not transfer: pro kilojoule cut-points, pro long-ride duration, and the idea that late-race power is trained only by late-race intervals.

Women are under-represented in the cyclist field studies. Main et al. (2026) shows carbohydrate availability changes the first threshold in trained women. Do not assume male professional record-power curves are their durability profile.

Open questions

  • A load-matched comparison of 2×20, 4×8, and 30/15 s for 20- and 40-minute power in trained amateurs at 7–12 h/week, with next-day performance as a fatigue outcome.
  • Early versus late interval placement, matched weekly load, with a fatigued time trial as the primary outcome.
  • Whether durability can improve while fresh CP stays flat, in already trained cyclists. Matomäki et al. (2023) saw something like this in untrained adults. Voet et al. (2025) saw both move.
  • Sex-specific agreement of FTP with CP/MLSS, and sex-specific durability after the same kJ·kg⁻¹ and the same relative intensity.
  • The minimum long-ride duration that changes subsequent fatigued performance when weekly hours are fixed.
  • Whether over-unders outperform steady work near CP for either FTP or durability.
  • How much of the CP drop is carbohydrate availability, muscle damage, falling economy, or falling VO₂max, in the same athletes, and which fraction is trainable.
  • A cyclist version of Talsnes et al. (2024): one long ride versus two shorter rides, then a multi-week performance outcome rather than next-morning soreness.

Search and access note

Europe PMC REST searches and PMC full text were used throughout. Web search was not available at the start of this investigation and was restored on 2026-10-02. It was then used for original-paper discovery and for 2025–2026 update scans (Voet 2025, Norte 2026, Matomäki 2023, Talsnes 2024, Mølmen 2025, Cove 2025 identification, and confirmation that sweet-spot and over-under recipe trials were not surfacing). Cove et al. (2025) remains abstract-only here. The polarised/non-polarised operationalization (interval duration under 5 minutes, not measured 80/20 or pyramidal zones) is taken from the distribution brief's full-text reading and was not re-extracted. The 2025 heart-rate time-in-zone individual-participant-data network meta-analysis is likewise not re-retrieved; the intensity-distribution brief holds that PDF. Snippets were not treated as results. Norte et al. (2026) and Voet et al. (2025) were re-read from PMC XML and the publisher PDF. Papers marked abstract-only in the capsules were not given unrecovered sample sizes or confidence intervals.

Publications with a retrieved date after 2026-10-02 were not used. Norte et al. (first publication 2026-06-01) and Main et al. (2026-06-02) fall inside the window. Sánchez-Jiménez et al. first appeared 2025-07-04; its journal issue is later. Its search closed in September 2024, so it is not a complete 2026 review.

Conflicts of interest were not systematically extracted from closed papers. Jones and Kirby (2025) report no specific funding for that review.

References

Aird TP, Davies RW, Carson BP. Effects of fasted vs fed-state exercise on performance and post-exercise metabolism: a systematic review and meta-analysis. Scand J Med Sci Sports. 2018. doi:10.1111/sms.13054. PMID 29315892. Abstract.

Barsumyan A, Soost C, Burchard R. Enhanced durability predicts success in amateur road cycling: evidence of power output declines. Front Sports Act Living. 2025a. doi:10.3389/fspor.2025.1530162. PMID 40438340. PMC12116599. Full text.

Barsumyan A, Soost C, Shyla R, Graw JA, Bliemel C, Burchard R. Durability as an independent parameter of endurance performance in cycling. BMC Sports Sci Med Rehabil. 2025b. doi:10.1186/s13102-025-01238-8. PMID 40640875. PMC12243304. Full text.

Borszcz FK, Tramontin AF, Bossi AH, Carminatti LJ, Costa VP. Functional threshold power in cyclists: validity of the concept and physiological responses. Int J Sports Med. 2018. doi:10.1055/s-0044-101546. PMID 29801189. Abstract retrieved via PubMed on 2026-10-02. Authors report no conflict of interest.

Borszcz FK, Tramontin AF, Costa VP. Is the functional threshold power interchangeable with the maximal lactate steady state in trained cyclists? Int J Sports Physiol Perform. 2019. doi:10.1123/ijspp.2018-0572. PMID 30676826. Abstract.

Caen K, Poole DC, Vanhatalo A, Jones AM. Critical power and maximal lactate steady state in cycling: “watts” the difference? Sports Med. 2024. doi:10.1007/s40279-024-02075-4. PMID 39196486. Abstract.

Chorley A, Lamb KL. The application of critical power, the work capacity above critical power (W′), and its reconstitution: a narrative review. Sports. 2020. doi:10.3390/sports8090123. PMID 32899777. PMC7552657. Abstract; open full text not re-read in full for this brief.

Clark IE, et al. Dynamics of the power-duration relationship during prolonged endurance exercise and influence of carbohydrate ingestion. J Appl Physiol. 2019. doi:10.1152/japplphysiol.00207.2019. PMID 31295069. Abstract.

Clark IE, et al. Changes in the power-duration relationship following prolonged exercise: estimation using conventional and all-out protocols and relationship with muscle glycogen. Am J Physiol Regul Integr Comp Physiol. 2019. doi:10.1152/ajpregu.00031.2019. PMID 30995104. Abstract.

Cove B, Chalmers S, Nelson MJ, Anderson M, Bennett H. The effect of training distribution, duration, and volume on VO₂max and performance in trained cyclists: a systematic review and meta-analysis. J Sci Med Sport. 2025. doi:10.1016/j.jsams.2024.12.005. PMID 39788807. Abstract only in this brief; first indexed date 2024-12-20. Do not read the polarised/non-polarised contrast as measured 80/20 versus pyramidal time in zone. The distribution brief's full text found the split used interval duration under 5 minutes. A 2025 heart-rate time-in-zone individual-participant-data network meta-analysis is the better polarised-versus-pyramidal source and was not re-extracted here.

Engelke P, et al. Equivalence between FTP and RCP: assessment of association, agreement and sex effect. Clinics. 2026. doi:10.1016/j.clinsp.2026.100977. PMID 42060964. PMC13139968. Abstract; open paper not re-read beyond the abstract.

Galán-Rioja MÁ, González-Mohíno F, Poole DC, González-Ravé JM. Relative proximity of critical power and metabolic/ventilatory thresholds: systematic review and meta-analysis. Sports Med. 2020. doi:10.1007/s40279-020-01314-8. PMID 32613479. Abstract.

Gallo G, et al. Power road-derived physical performance parameters in junior, under-23, and professional road cycling climbers. Int J Sports Physiol Perform. 2022. doi:10.1123/ijspp.2021-0554. PMID 35483700. Abstract.

Gejl KD, et al. No superior adaptations to carbohydrate periodization in elite endurance athletes. Med Sci Sports Exerc. 2017. doi:10.1249/MSS.0000000000001377. PMID 28723843. Abstract.

Hamilton K, et al. Durability of the moderate-to-heavy-intensity transition is related to the effects of prolonged exercise on severe-intensity performance. Eur J Appl Physiol. 2024. doi:10.1007/s00421-024-05459-6. PMID 38546844. Abstract.

Helgerud J, et al. Aerobic high-intensity intervals improve VO₂max more than moderate training. Med Sci Sports Exerc. 2007. doi:10.1249/mss.0b013e3180304570. PMID 17414804. Abstract.

Hunter B, Maunder E, Jones AM, Gallo G, Muniz-Pumares D. Durability as an index of endurance exercise performance: methodological considerations. Exp Physiol. 2025. doi:10.1113/EP092120. PMID 40150840. PMC12576026. Full text skimmed for methods cautions.

Impey SG, et al. Fuel for the work required: a theoretical framework for carbohydrate periodization and the glycogen threshold hypothesis. Sports Med. 2018. doi:10.1007/s40279-018-0867-7. PMID 29453741. Abstract.

Jones AM, Burnley M, Black MI, Poole DC, Vanhatalo A. The maximal metabolic steady state: redefining the ‘gold standard’. Physiol Rep. 2019. doi:10.14814/phy2.14098. PMID 31124324. PMC6533178. Full text.

Jones AM. The fourth dimension: physiological resilience as an independent determinant of endurance exercise performance. J Physiol. 2024. doi:10.1113/JP284205. PMID 37606604. Abstract.

Jones AM, Kirby BS. Physiological resilience: what is it and how might it be trained? Scand J Med Sci Sports. 2025. doi:10.1111/sms.70032. PMID 40024804. PMC11872681. Full text.

Kjøsen Talsnes R, Torvik PØ, Skovereng K, Sandbakk Ø. Comparison of acute physiological responses between one long and two short sessions of moderate-intensity training in endurance athletes. Front Physiol. 2024. doi:10.3389/fphys.2024.1428536. PMID 39139482. Abstract.

Laursen PB, Shing CM, Peake JM, Coombes JS, Jenkins DG. Interval training program optimization in highly trained endurance cyclists. Med Sci Sports Exerc. 2002. doi:10.1097/00005768-200211000-00017. PMID 12439086. Abstract.

MacInnis MJ, Thomas ACQ, Phillips SM. The reliability of 4-minute and 20-minute time trials and their relationships to functional threshold power in trained cyclists. Int J Sports Physiol Perform. 2019. doi:10.1123/ijspp.2018-0100. PMID 29809063. Abstract.

Main E, et al. Low carbohydrate availability reduces power output at the moderate-to-heavy transition, impairs efficiency, and increases median power frequency during cycling in females. Eur J Appl Physiol. 2026. doi:10.1007/s00421-026-06260-3. PMID 42228164. Abstract.

Marquet LA, et al. Enhanced endurance performance by periodization of carbohydrate intake: “sleep low” strategy. Med Sci Sports Exerc. 2016. doi:10.1249/MSS.0000000000000823. PMID 26741119. Abstract.

Mateo-March M, et al. The record power profile of male professional cyclists: fatigue matters. Int J Sports Physiol Perform. 2022. doi:10.1123/ijspp.2021-0403. PMID 35240578. Abstract.

Mateo-March M, et al. Is all work the same? Performance after accumulated work of differing intensities in male professional cyclists. J Sci Med Sport. 2024. doi:10.1016/j.jsams.2024.03.005. PMID 38604818. Abstract.

Mateo-March M, et al. Reliability of the durability concept in professional cyclists: a field-based study. Int J Sports Med. 2025. doi:10.1055/a-2555-8961. PMID 40373793. Abstract.

Matomäki P, et al. Durability is improved by both low and high intensity endurance training. Front Physiol. 2023. doi:10.3389/fphys.2023.1128111. PMID 36875044. PMC9977827. Full text.

Maunder E, Seiler S, Mildenhall MJ, Kilding AE, Plews DJ. The importance of ‘durability’ in the physiological profiling of endurance athletes. Sports Med. 2021. doi:10.1007/s40279-021-01459-0. PMID 33886100. Abstract.

McGrath E, Mahony N, Fleming N, Donne B. Is the FTP test a reliable, reproducible and functional assessment tool in highly-trained athletes? Int J Exerc Sci. 2019. doi:10.70252/RQOO7391. PMID 31839854. Abstract.

Morgan PT, Black MI, Bailey SJ, Jones AM, Vanhatalo A. Road cycle TT performance: relationship to the power-duration model and association with FTP. J Sports Sci. 2019. doi:10.1080/02640414.2018.1535772. PMID 30387374. Abstract.

Mølmen KS, et al. A moderate-intensity interval training block improves endurance performance in well-trained cyclists. Med Sci Sports Exerc. 2025. doi:10.1249/MSS.0000000000003706. PMID 40101160. Abstract.

Neal CM, et al. Six weeks of a polarized training-intensity distribution leads to greater physiological and performance adaptations than a threshold model in trained cyclists. J Appl Physiol. 2013. doi:10.1152/japplphysiol.00652.2012. PMID 23264537. Abstract.

Norte BR, et al. Graded carbohydrate ingestion up to 120 g·h⁻¹ attenuates the reduction in critical power following 3 h of moderate-intensity exercise in a dose-dependent manner. Scand J Med Sci Sports. 2026. doi:10.1111/sms.70326. PMID 42322010. PMC13282445. Full text.

Ørtenblad N, Zachariassen M, Nielsen J, Gejl KD. Substrate utilization and durability during prolonged intermittent exercise in elite road cyclists. Eur J Appl Physiol. 2024. doi:10.1007/s00421-024-05437-y. PMID 38441690. Abstract.

Peeters WM, Barrett M, Podlogar T. What is a cycling race simulation anyway: a review on protocols to assess durability in cycling. Eur J Appl Physiol. 2025. doi:10.1007/s00421-025-05725-1. PMID 39953333. PMC12174182. Full text.

Poole DC, Burnley M, Vanhatalo A, Rossiter HB, Jones AM. Critical power: an important fatigue threshold in exercise physiology. Med Sci Sports Exerc. 2016. doi:10.1249/MSS.0000000000000939. PMID 27031742. Abstract.

Rosenblat MA, Perrotta AS, Vicenzino B. Polarized vs. threshold training intensity distribution on endurance sport performance: a systematic review and meta-analysis of randomized controlled trials. J Strength Cond Res. 2019. doi:10.1519/JSC.0000000000002618. PMID 29863593. Abstract.

Rønnestad BR, Hansen EA, Raastad T. Effect of heavy strength training on thigh muscle cross-sectional area, performance determinants, and performance in well-trained cyclists. Eur J Appl Physiol. 2010. doi:10.1007/s00421-009-1307-z. PMID 19960350. Abstract.

Rønnestad BR, Hansen EA, Raastad T. Strength training improves 5-min all-out performance following 185 min of cycling. Scand J Med Sci Sports. 2011. doi:10.1111/j.1600-0838.2009.01035.x. PMID 19903319. Abstract.

Rønnestad BR, Hansen J, Vegge G, Tønnessen E, Slettaløkken G. Short intervals induce superior training adaptations compared with long intervals in cyclists — an effort-matched approach. Scand J Med Sci Sports. 2015. doi:10.1111/sms.12165. PMID 24382021. Abstract.

Rønnestad BR, Hansen J, Nygaard H, Lundby C. Superior performance improvements in elite cyclists following short-interval vs effort-matched long-interval training. Scand J Med Sci Sports. 2020. doi:10.1111/sms.13627. PMID 31977120. Abstract.

Sánchez-Jiménez JL, et al. Is intensity the most important factor in determining the amount of prior work accumulated that affects cyclists’ acute durability? A systematic review. Eur J Appl Physiol. 2025. doi:10.1007/s00421-025-05885-0. PMID 40613880. PMC12881052. Full text. First publication 2025-07-04.

Seiler S. What is best practice for training intensity and duration distribution in endurance athletes? Int J Sports Physiol Perform. 2010. doi:10.1123/ijspp.5.3.276. PMID 20861519. Abstract.

Seiler S, Jøranson K, Olesen BV, Hetlelid KJ. Adaptations to aerobic interval training: interactive effects of exercise intensity and total work duration. Scand J Med Sci Sports. 2013. doi:10.1111/j.1600-0838.2011.01351.x. PMID 21812820. Abstract.

Sitko S, et al. What is “zone 2 training”?: experts’ viewpoint on definition, training methods, and expected adaptations. Int J Sports Physiol Perform. 2025. doi:10.1123/ijspp.2024-0303. PMID 40010355. Abstract.

Spragg J, Leo P, Swart J. The relationship between physiological characteristics and durability in male professional cyclists. Med Sci Sports Exerc. 2023a. doi:10.1249/MSS.0000000000003024. PMID 35977108. Abstract.

Spragg J, Leo P, Swart J. The relationship between training characteristics and durability in professional cyclists across a competitive season. Eur J Sport Sci. 2023b. doi:10.1080/17461391.2022.2049886. PMID 35239466. Abstract.

Stepto NK, Hawley JA, Dennis SC, Hopkins WG. Effects of different interval-training programs on cycling time-trial performance. Med Sci Sports Exerc. 1999. doi:10.1097/00005768-199905000-00018. PMID 10331896. Abstract.

Stevenson JD, Kilding AE, Plews DJ, Maunder E. Prolonged cycling reduces power output at the moderate-to-heavy intensity transition. Eur J Appl Physiol. 2022. doi:10.1007/s00421-022-05036-9. PMID 36127418. Abstract.

Stöggl T, Sperlich B. Polarized training has greater impact on key endurance variables than threshold, high intensity, or high volume training. Front Physiol. 2014. doi:10.3389/fphys.2014.00033. PMID 24550842. Abstract. Cited only as background to the polarised-versus-threshold comparison; the cyclist-specific trials carry the programming claim. Cove et al. (2025) is not used here as a polarised-versus-pyramidal result.

Sylta Ø, et al. The effect of different high-intensity periodization models on endurance adaptations. Med Sci Sports Exerc. 2016. doi:10.1249/MSS.0000000000001007. PMID 27300278. Abstract.

Valenzuela PL, Morales JS, Foster C, Lucia A, de la Villa P. Is the functional threshold power a valid surrogate of the lactate threshold? Int J Sports Physiol Perform. 2018. doi:10.1123/ijspp.2018-0008. PMID 29745765. Abstract.

Voet JG, Lamberts RP, Viribay A, de Koning JJ, van Erp T. Durability and underlying physiological factors: how do they change throughout a cycling season in semiprofessional cyclists? Int J Sports Physiol Perform. 2024. doi:10.1123/ijspp.2023-0543. PMID 38871342. Abstract.

Voet JG, van Erp T, Viribay A, de Koning JJ, Lamberts RP. Training characteristics related to (the changes in) durability in semiprofessional cyclists. Int J Sports Physiol Perform. 2025. doi:10.1123/ijspp.2024-0321. PMID 40054482. Publisher PDF.

Zanini M, Folland JP, Blagrove RC. The effect of 90 and 120 min of running on the determinants of endurance performance in well-trained male marathon runners. Scand J Med Sci Sports. 2025. doi:10.1111/sms.70076. PMID 40375575. Abstract.

Zanini M, Jones AM, Nybo L. Defining physiological resilience and durability in the context of endurance performance modeling. J Appl Physiol. 2025. doi:10.1152/japplphysiol.00671.2025. PMID 41396096. Metadata only: no abstract was available in Europe PMC on the check date, so no findings are used.