REF-1239 · Sánchez-Jiménez et al., 2026 · European Journal of Applied Physiology · Paper · PubMed · Free full text

Does hard riding tire me more than the same kJ easy?

Yes, mostly in short efforts: a review of 21 cycling studies found work above CP cut later power more per kJ; 12-min power and CP held in two trials. No pooling; mostly male pros.

B+moderate evidenceMeasured on highly trained to world-class athletes

Sánchez-Jiménez et al. asked how the intensity of prior work affects how much work it takes to reduce a cyclist's power within a single session, and whether kilojoules are a fair fatigue measure.

They searched Web of Science, Medline and Scopus (18 September 2024; OSF registration osf.io/kcg53) for studies measuring power fresh and after fatigue induced in the same session, with prior work in kJ or another load metric, and included 21 (14 from the search, 7 from reference lists and expert recommendation) with 585 participants.

Designs ranged from mining professionals' race and training files to laboratory and field protocols from a 38-min laboratory ride to a 6-h race simulation.

The authors report that efforts above CP reduced power by 10-20% with 2.5-15 kJ/kg of work, whereas work below CP typically yielded under 5% even at higher volumes; that higher-level riders needed more work above CP before their record powers fell; and that CP itself can miss fatigue that shows in shorter efforts.

They conclude that intensity is the most critical factor and that kJ alone is insufficient.

Our reading: the direction is consistent across the work-matched comparisons, and it adds field data to the single trial the base cited (REF-193, which is one of the 21); but the headline numbers are a narrative span across unlike designs, in the two controlled trials behind the headline (Spragg 2024, with less work in the hard arm, and the work-matched Barranco-Gil 2024) the extra cost of intensity showed only in efforts of 3 min or shorter, while the time-matched Leo 2022 pilot (9 riders, about 237 kJ more in the hard arm) found it in 12-min power and the work-matched MacDougall 2024 (16 recreational riders) in a roughly 9-min effort to failure, and the review authors wrote several of the studies they synthesise.

Protocol

Question (PICO as stated): which amount of fatigue impacts cyclists' performance assessed through power output reduction, and is mechanical work the most effective variable to monitor fatigue?
Search: Web of Science, Medline (PubMed) and Scopus on 18 September 2024, English or Spanish; terms durability or fatigue, cycling or cyclist, load or work or workload, power output. 913 records, 511 after duplicates, 27 read in full, 14 included, plus 7 from other sources: 21 studies, n = 585 participants.
Inclusion: power measured fresh and after a fatiguing cycling session in the same session, with prior fatigue quantified (kJ, TSS, TRIMP or sRPE) or calculable; multi-day designs such as Grand Tour simulations excluded by design.
Extraction and bias: two authors in parallel, a third for disagreements; figures digitised where needed. Newcastle-Ottawa Scale for 8 observational studies (three scored 7, five scored 8), ROBINS-I for 7 quasi-experimental studies (71.4% low overall risk), RoB 2 for 6 randomised studies (83.3% low overall risk).
Synthesis: tabular and narrative; no meta-analysis, no effect sizes pooled, no certainty grading.
Funding and interests: open-access funding from Universidad de Granada/CBUA; predoctoral fellowships from the Spanish Ministry of Universities to two authors; no conflicts declared. Review authors are authors of at least six included studies (Mateo-March 2022a and 2024, Gallo 2022, Muriel 2022, Leo 2024, Sanchez-Jimenez 2023; our count from the reference list); this is not declared as a conflict.

Key Results

Work-matched or intensity-contrasted studies in the review (values from its Table 1; W/kg changes from fresh, several digitised from figures by the reviewers):

StudyRidersContrastResult
Spragg 2024 (REF-193)14 elite/international men1985 ± 242 kJ below 70% CP vs 1878 ± 340 kJ including 5 × 8 min at 105-110% CP (less work in the hard arm, significantly less per REF-193)1-s −0.56 vs −1.57; 15-s −0.29 vs −1.47; 3-min −0.17 in both; 12-min −0.05 vs 0; CP +0.007 vs +0.06
Barranco-Gil 202412 junior men15 kJ/kg at 65% CP vs 3-min intervals at ~115% CP (TSS 92 vs 134)2-min power fell after the intervals (per the review's text); 5-min, 12-min and CP did not (CP −0.0003 vs +0.03)
Leo 20229 ProTeam men150 min moderate (~1750 kJ) vs race simulation (~2000 kJ)12-min power −0.46 after the race simulation; 0.5 W/kg between protocols
Mateo-March 202417 WorldTour men, field fileswork above vs below CP, up to 7.5 kJ/kgCP −0.99 vs −0.07; 20-min −1.17 vs −0.09; 30-s −2.67 vs −0.22
Leo 202411 under-23, 13 ProTeam, 24 WorldTour men, field files2.5-7.5 kJ/kg above CPat least 2.5 kJ/kg above CP needed in all; under-23 declined after 2.5-5.0, ProTeam and WorldTour after 5.0-7.5
MacDougall 20248 men, 8 women, recreationally active10 × 2 min at 80% peak power vs 38 min at 54%, ~293 vs ~297 kJtime to task failure at 80% peak power: 548 s fresh, 82 s after intervals vs 208 s after steady riding
Fullerton 20216 men, 5 womenprior exercise above vs below MLSSgreater decline above MLSS
Headline (Discussion): efforts above CP reduced power by 10-20% with 2.5-15 kJ/kg of work; below CP, typically under 5% even at higher volumes (citing Mateo-March 2024, Spragg 2024, Barranco-Gil 2024).
Work alone (kJ-only studies): in 112 professionals record powers fell 1.6-3.0% after 15 kJ/kg and 6.0-9.7% after 45 kJ/kg, more in ProTeam than WorldTour riders (REF-192); 12 national elite riders lost 28 W of 7-min power over a 6-h race simulation, 14 W of it in the last 2 h (REF-474); 12 elite riders lost 0.55 W/kg of 6-min power after 4 h of intermittent riding (REF-144).
Level: WorldTour riders needed more work before their power fell than ProTeam, under-23 or junior riders; 10 trained/developmental riders (best 20 min 4.9 W/kg) lost 9% of 1-min, 5.9% of 5-min and 4.1% of 20-min power after only 3.5 kJ/kg of hard intervals.
CP as the outcome: two studies found CP tracked the decline and two did not (Spragg 2024, Barranco-Gil 2024); the authors caution that CP may not capture fatigue.
Numbers that do not add up:
The headline 10-20% cannot be traced, for efforts longer than 3 min, to two of its three cited studies, the controlled trials (Spragg 2024, less work in the hard arm; Barranco-Gil 2024, work-matched): in Spragg 2024 12-min power and CP did not fall after the hard work, and in Barranco-Gil 2024 5-min, 12-min and CP did not; the falls there were in efforts of 3 min or shorter.The 10-20% span for longer efforts rests on field files (Mateo-March 2024: CP −0.99 W/kg); the review gives no fresh values, so the percentages cannot be recomputed from it.
Van Erp and Lamberts 2022 is tabulated as "N = 14 (M) Professional" and counted as 14 men, but its title in the reference list is about female professional cycling; the women in the review are therefore probably 46 of 585 rather than 32 (our count).
"Four studies focused on laboratory testing" is followed by a list of five, and "four studies analysed race and training data" is also followed by five (only five make the stated eight database studies add up).

Mechanisms

Measured (in the included studies): after hard vs easy protocols of similar or smaller work, the high-intensity version cut 1-s and 15-s power more than steady riding (Spragg 2024, where 3-min power fell 0.17 W/kg after both) and 2-min power fell after the intervals (Barranco-Gil 2024); 12-min power and CP did not change in these two controlled trials, whereas a roughly 9-min effort to failure at 80% of peak power fell far more after work-matched intervals (MacDougall 2024, recreational riders), while in professionals' field files power over 5-20 min and modelled CP fell after a few kJ/kg above CP.
Authors' interpretation: high-intensity work depletes glycogen and fatigues fast-twitch fibres and the neuromuscular system, so short efforts suffer most; a fatigued fall in power is driven mainly by lower cadence rather than torque; CP, modelled from maximal efforts of about 3-12 min, may miss submaximal fatigue. They propose intensity-weighted load measures (kJ normalised to %CP or FTP, TSS, TRIMP, their Power Profile Index), none validated for durability.
Inference: a fatigued power profile read on a kJ/kg axis (REF-479) mixes riders and days whose kilojoules were done at very different intensities; the same kJ/kg after a surge-filled race and after a steady ride are not the same dose.
Inference: the review cannot separate intensity from glycogen use or fuelling, which it did not extract; its authors point to carbohydrate intake as a moderator (REF-1098).

Practical Application

Does hard riding tire me more than the same kJ easy? Yes, mostly in short efforts: a review of 21 cycling studies found work above CP cut later power more per kJ; 12-min power and CP held in two trials. No pooling; mostly male pros.
Count the hard work, not only the kilojoules: when reading a long ride or race for fatigue, the minutes above CP matter more than the total kJ. This supports SYNTH-32's "hard prior work costs more than a lot of easy work", which rested on one trial of 14 professionals (REF-193), now with field data in professionals behind it; the size of the extra cost is not pooled.
Expect short efforts to go first: in the two controlled trials behind the headline, sprint to 3-min power fell after hard work while 12-min power and CP held; in recreational riders, though, work-matched intervals cut a roughly 9-min effort to failure far more than steady riding (MacDougall 2024). A rider who must attack late needs that capacity protected, not only threshold power (inference from the pattern).
Lower-level riders fade sooner: junior and under-23 riders, and trained amateurs in one study, lost power after less work than WorldTour riders; amateurs should not be given professional kJ/kg benchmarks.
Training on it is not tested here: the review is about acute fatigue. The authors' suggestion of race-like hard efforts (for example 5 × 8 min at about 105% CP) to build tolerance is their proposal, not a training trial (SYNTH-32 §5 lists what has been trained).
Fuel and environment were not controlled across studies (REF-1098); an under-fuelled rider will also fade sooner (SYNTH-32).

Limitations

No pooling, mixed designs. Observational file-mining, non-randomised and randomised experiments are summarised side by side; there is no effect size, no heterogeneity estimate and no certainty grading, so "intensity is the most critical factor" is the authors' narrative conclusion.
Search sensitivity: about 30% of the included studies came from manual screening, which the authors attribute to inconsistent terminology (durability, fatigue resistance, resilience).
The headline range spans different outcomes, protocols and levels and is not consistent with the long-effort results of two of its three cited studies.
Who: almost all male road cyclists, most of them professional or under-23; recreational riders and women are few (and 14 female professionals are miscounted as men).
Author overlap: review authors wrote at least six of the 21 studies, including several of the field analyses behind the headline; not declared.
Acute only: single-session fatigue; multi-day racing and training adaptations are outside its scope, and fuelling was not extracted.
How this entry was checked and changed
Why this level

systematic search and bias rating but no pooling; 21 heterogeneous studies (8 observational analyses of race and training files, 7 non-randomised and 6 randomised within-subject experiments, the experiments with 9-16 riders each apart from a national-team testing database (119 riders) and Leo 2024 (48 riders' files with laboratory CP), which the review leaves out of its eight database studies); about a third of the studies came from outside the database search; the 10-20% headline is a narrative range, not a computed estimate; review authors wrote at least six of the included studies; read from the full text (open access, CC BY 4.0)

Full text (checked 2026-10-26)
Of 913 initial studies, 511 unique studies remained after removing duplicates; 27 were left for full-text review, of which 14 met inclusion criteria, and an additional 7 studies were identified through other sources (full text)
A total of 585 participants were included in the studies analysed in this review (full text)
Efforts above CP reduced PO by 10–20% with 2.5–15 kJ kg–1 of work, whereas low-to-moderate intensity efforts below CP typically yielded < 5% reductions even at higher volumes (full text)
For ROBINS-I, 71.4% of studies had a low overall risk; for RoB-2, 83.3% of studies had low overall risk (full text)
Leo et al. (2024): all cyclists required at least 2.5 kJ kg–1 above CP to exhibit a significant decline in MMP; U23 cyclists declined after 2.5–5.0, PT and WT cyclists only after 5.0–7.5 kJ kg–1 above CP (full text)
Spragg et al. (2024) found that high-intensity work decreased PO in shorter efforts (1-s, 15-s, and 3-min) but did not significantly affect 12-min TT performance; CP did not differ between fresh and fatigued states (full text)
Sanchez-Jimenez et al. (2023): performance reductions after an accumulated workload of only 3.5 ± 0.2 kJ kg–1, with decrements of 9% (1-min), 5.9% (5-min), and 4.1% (20-min) (full text)
Approximately 30% of the included studies were identified through manual screening (full text)
Van Erp and Lamberts (2022) is tabulated as N = 14 (M) Professional, while its reference title reads performance characteristics of TOP5 versus NOT-TOP5 races in female professional cycling (full text)
Predoctoral fellowships from the Spanish Ministry of Universities (FPU20/00611 and FPU22/02694); the authors declare no conflicts of interest (full text)