REF-741 · Hays A, Devys S, Bertin D, Marquet L-A, Brisswalter J — 2018 · Frontiers in Physiology · Paper
A modern Olympic cross-country MTB race is run at 87% VO₂max with 37% of the time above VT2 and 25% above maximal aerobic power, delivered as bursts of 5–30 s
And time-in-zone by heart rate or power does not track time-in-zone by oxygen uptake
Protocol
Lab: VO₂ and HR expressed as %VO₂max and %HRmax and divided into three zones — Z1 (below VT1), Z2 (VT1–VT2), Z3 (VT2–VO₂max); for power an additional fourth zone above maximal aerobic power (MAP).
Field: one simulated and one official XCO race on a cycling-accredited track; measurements were taken in the simulated race only, the official race giving the reference run time (corrected 2026-10-12, full text). Full text (not in the abstract): portable gas analyser (Cosmed K5), heart-rate monitor and power meter; capillary blood lactate, pH and bicarbonate sampled pre-exercise and after each lap.
Key Findings
| Measure | Value |
|---|---|
| Mean intensity | 87% VO₂max |
| Time above VT2 | 37.0 ± 17.9% of the race |
| Time above MAP (power) | 25% of the race |
| Burst duration above MAP | most periods beyond MAP lasted 5–30 s, with low-intensity periods between the short repeated bursts |
| Burst structure, full text (not in the abstract) | 34.5 ± 8.6% of above-MAP time in 5–10 s efforts; ~22 actions of 1–5 s per lap; most above-MAP actions last under 10 s |
| Blood lactate, full text (not in the abstract) | 6.5 ± 0.9 mmol/L mean over the race (1.9 ± 0.5 pre-exercise), differing significantly between laps |
The first lap is very high intensity; subsequent laps are more aerobic — a fast-start profile.
Zone shares by %VO₂max were uncorrelated with zone shares by %HRmax (r = −0.01) or %MAP (r = −0.02) (full text; not in the abstract): heart rate and power each describe a different race from the one the lungs ran. Inference from the authors' discussion: VO₂ stays high through technical sections where power is low.
The authors' conclusion: the current XCO format is acyclical and intermittent exercise comparable to high-intensity team sports, and VO₂ values should be combined with the usual HR and power data when analysing XCO performance. Full text: the comparison is to soccer or basketball, and the technical bursts involve static and eccentric upper-body work.
Practical Application
An XCO race is 87% VO₂max on average with 37.0% of the time above VT2 and 25% above MAP, delivered as 5–30 s bursts with low-intensity periods between them (16 national/international riders, measured in a simulated race on an official race track (corrected 2026-10-12, full text)) — train off-road racers for repeatability of short bursts above MAP on a threshold-tempo base, with only seconds between them: 30/15 sets (REF-045/REF-046), bursts inside threshold blocks, simulated laps.Long steady VO₂max intervals raise the ceiling; they do not rehearse this pattern.
Read an off-road race file by the count and length of supra-MAP efforts and by time above MAP, not by average or normalised power, and do not judge race intensity from heart-rate zones alone — here HR-zone time and VO₂-zone time were unrelated (full text; not in the abstract).
Inference, not tested: for cyclocross the burst structure transfers and the HR side is even flatter (REF-740); the aerobic base underneath is what road and gravel volume builds.
Limitations
XCO, not cyclocross: races of about 1 h 30 with descents (full text; not in the abstract); Not from this paper: cyclocross is shorter and has no descents (see REF-740), so the recovery between bursts is smaller in cross.
Young national/international riders (Tier 3-4); burst intensities in W/kg will not transfer to amateurs, the pattern will.
Effort counts depend on the threshold chosen (MAP) and on sampling; the systematic review in REF-742 collects widely varying counts across studies.
The 25% above MAP in the abstract and the per-zone percentages in the full text differ slightly (28.2 ± 8.1% of race time in P4 in the full text; the 22 ± 6% figure is Stapelfeldt et al. 2004 for the old format, cited by the authors (corrected 2026-10-12, full text)); treat "about a quarter of race time above MAP" as the figure.
How this entry was checked and changed
Content correction 2026-09-27
Rider characteristics (age 17.4, 15–23; VO₂max 64.6; MAP 5.2 W/kg), burst breakdown (34.5 ± 8.6% in 5–10 s; ~22 actions of 1–5 s per lap), lactate 6.5 ± 0.9 and the r = −0.01 / −0.02 correlations are not in the abstract; each is now labelled full text after checking against the PMC full text (all present).
"roughly 130–150% of FTP" -> deleted (house number, not in the paper); "40–60 min" cyclocross duration and "18 per lap to 334 per race" effort counts -> deleted (not from this paper; REF-742 is live-only and could not be checked).
"lactate highest after lap 1 and falling by lap 3" -> "differing significantly between laps" (what the full text states); "(most under 10 s)" moved from the title to the labelled full-text row.
Full text (checked 2026-10-12)
Correction: the entry says gas, HR and power were recorded in "one simulated and one official XCO race" and calls this "the first study to measure oxygen uptake during a real XCO race", but the full text states the official competition "was used as reference for run time" and gas exchange, HR and power were recorded during the simulated race on the same track (full text).
Correction: the entry attributes "22 ± 6% in P4 in one analysis" to this paper, but in the full text 22 ± 6% is the P4 share reported by Stapelfeldt et al. (2004) for the old format; this study found 28.2 ± 8.1% of race time at P4 (full text).
The simulated race was three laps of a 5.10 km track (215 m cumulative altitude difference) with a 30 s rest between laps, two cyclists of similar level per run, and lasted 64 ± 1.5 min versus 66 ± 2 min for the official race (full text).
Time above VT2 by VO2 fell from 48.8 ± 15.1% in lap 1 to 34.7 ± 15.6% in lap 2 and 27.4 ± 16.0% in lap 3, while time above MAP fell from 32.8 ± 8.2% to 27.9 ± 7.9% and 24.0 ± 8.2% (full text, Table 1 and Table 2).
By heart rate, 15.1 ± 2.1% of the race was below VT1, 52.1 ± 1.5% between VT1 and VT2 and 32.8 ± 2.7% above VT2, with no significant difference between laps (full text).
Riders spent 19.2 ± 5.0% of the race not pedalling (NP) and 30.6% ± 9.4% in P1 below VT1 power (full text).
Blood pH fell from 7.4 ± 0.0 pre-exercise to 7.3 ± 0.0 and bicarbonate from 22.7 ± 1.7 to 14.7 ± 1.6 mmol-1 over the race, with the largest drop after the first lap (full text).
The 1–5 s category held n = 22.1 efforts on average (almost 50% of efforts above MAP); the full text does not state that this count is per lap (full text).