REF-746 · Mougin, Bailey, Jones, Joyner, Mears, Pearce & Zanini, 2026 · Sports Medicine, 56, 2047–2065 · Paper

Across 888 athletes from recreational to world class, V̇O₂max explained 65–76% of the between-athlete differences in power or speed at the lactate thresholds, economy 20–24%, and fractional utilisation only 4–11%

B+moderate evidenceMeasured on recreational to world-class athletesWritten before October 2026, being re-checked

Protocol

495 runners (105 women) and 393 cyclists (42 women) completed incremental exercise tests at one laboratory group (Loughborough, UK).

Measured: V̇O₂max (runners 56 mL·kg⁻¹·min⁻¹ / 3.94 L·min⁻¹ on average; cyclists 52 / 3.99), economy (runners 220 mL·kg⁻¹·km⁻¹; cyclists 14.7 mL·min⁻¹·W⁻¹), fractional utilisation of V̇O₂max at the lactate threshold (FU-LT: runners 78%, cyclists 70%) and at the lactate turnpoint (FU-LTP: runners 88%, cyclists 84%), and speed or power at LT (12.0 km·h⁻¹ / 190 W) and LTP (13.9 km·h⁻¹ / 240 W).

Single and multiple linear regressions related each determinant to the performance proxies (speed or power at LT and LTP) and quantified the relative contribution of each.

Key Findings

Speed or power at LT and LTP correlated strongly with V̇O₂max: R² = 0.65–0.77 (P < 0.001), and inversely with economy (R² = 0.24–0.26, P < 0.001).
Fractional utilisation was trivially related to threshold speed or power: R² ≤ 0.04 for FU-LT and ≤ 0.01 for FU-LTP.
Regression models built from the three determinants reproduced measured LT and LTP almost exactly (R² = 0.94–0.99), which is the Joyner model working as designed — the three factors do multiply into threshold performance.
Relative contribution to the performance proxies: V̇O₂max 65–76%, economy 20–24%, fractional utilisation 4–11%.
The authors' conclusion: V̇O₂max and economy together predict ~95% of speed or power at LT and LTP, "whilst the contribution of FU-LT or FU-LTP is limited in populations with heterogeneous characteristics".

Read the direction carefully — between athletes, not within one

The question the study answers is: when two athletes differ in threshold power, which determinant explains the difference? In a cohort spanning recreational to world class, the answer is overwhelmingly V̇O₂max. That is a statement about variance in a heterogeneous population.

It is not a statement that fractional utilisation cannot change within one athlete, and it says nothing about which lever moves fastest in training — a cross-sectional regression cannot see training.

Two things compress the fractional-utilisation term by construction. Its range across the cohort is narrow (FU-LTP 84–88% on average, bounded above by 100%), while V̇O₂max ranges over nearly threefold. And it is a ratio whose denominator is V̇O₂max, so once V̇O₂max is in the model, FU can only add what V̇O₂max has not already explained.

Both are real properties of physiology, not artefacts, but they mean "4–11%" is the share of BETWEEN-athlete variance, not a measure of how much FU matters to any given rider.

The newsletter that surfaced this paper (Knowledgeiswatt #191) reports it as "393 cyclists" — that is the cycling subgroup; the study is 888 individuals across both sports, and the runners showed the same pattern.

Practical Application

Between athletes, the gap to the next level is mostly V̇O₂max, then economy — this does not name one athlete's lever. A cross-section cannot see training. Not from this paper; the base's own practice: work above threshold is the starting bet for most amateurs (SYNTH-42). Building a plan around "raising fractional utilisation" as a separate target is low-yield at this level; between athletes, threshold power follows the ceiling.
The LT2/V̇O₂max ratio is a note, not a decision axis. Across this population a high ratio (the existing engine already well used) is far more common than a low one — that does not say it is true of the athlete in front of you, and a low ratio is not by itself a reason for a threshold block (inference, SYNTH-42).
At the top end the ordering changes. Once V̇O₂max is at the level's ceiling (the newsletter author's rule of thumb for male professional road cyclists is ~80 mL·kg⁻¹·min⁻¹, and that number is opinion, not from this paper), economy and fractional utilisation are what separate riders — REF-541 shows the same in world-class skiers. Do not carry the 65–76% figure into a homogeneous elite group.
Economy is the second lever and it is trainable — its 20–24% share is not small, and it is the one determinant that keeps improving after V̇O₂max plateaus.
Do not tell an athlete "V̇O₂max is all that matters". The correct sentence is that V̇O₂max sets the range within which economy and utilisation decide the outcome, and that the range is wide.

Limitations

Cross-sectional: explains variance between athletes, not change within one. No training intervention.
Performance proxies are laboratory LT and LTP speed or power, not race results; the leap from "power at LTP" to "endurance performance" is the Joyner model's own assumption.
A deliberately heterogeneous cohort (recreational to world class) maximises the V̇O₂max share; the same regression inside a homogeneous group would reallocate variance toward economy and FU.
Fractional utilisation is bounded and narrow by nature, and its denominator is V̇O₂max, so its independent contribution is structurally small in this design.
Single laboratory group and incremental protocols; the split between LT and LTP definitions follows that group's conventions.
Women are 17% of the cohort; no sex-specific breakdown is reported in the abstract.
How this entry was checked and changed
Content correction 2026-09-27
Source note (truth pass): the numbers in this entry were taken from the paper's full text when it was ingested; many of them are not in the abstract. They were not re-checked against the full text on 2026-09-27.
Cross-reference descriptions checked against the cited entries' current content; wrong years, wrong entries and claims the cited entry does not make were corrected.
2026-10-02: aligned with the rewritten SYNTH-42 (missing stimulus, one lever per block, level C+): the first two Practical bullets read the between-athlete regression as a rule for choosing one athlete's lever ("the engine is the lever at Tier 1-3", "read the ratio, the ceiling is the limiter"), which this entry's own Limitations and SYNTH-42 rule out; reworded to the between-athlete finding, with the starting bet labelled practice and the ratio a note. No number changed.
Full text (checked 2026-10-12)
LT was defined as the first rise in blood lactate from baseline (+ 0.4–0.5 mmol/L) and LTP as a rapid and sustained increase (+ 1.5 mmol/L) between two subsequent stages of a 4-min-stage incremental test, followed after 15–20 min of rest by a separate test to task failure for V̇O₂max (full text).
By sex, V̇O₂max contributed 80.7% (females) vs 70.0% (males) to running speed at LT and 85.9% vs 73.6% to speed at LTP, while running economy contributed 14.8% vs 22.3% at LT (full text, Table 5).
In cycling, FULT contributed 11.0% to power at LT overall (12.9% in males, 16.4% in females) but FULTP only 5.8% to power at LTP; in female cyclists power at LT correlated with FULT (R2 = 0.27) (full text, Table 5).
The LT/LTP ratio averaged 86 (3)% in runners and 79 (4)% in cyclists and was not correlated with speed or power at LTP (running R2 = 0.001; cycling R2 = 0.025) (full text, Table 1).
Heart rate at LT averaged 86% of HRmax in runners and 80% in cyclists, and at LTP 94% and 89%, with no correlation between threshold speed or power and %HRmax at the threshold (full text, Table 1).
Ten-fold cross-validation gave R2 of 0.94 to 0.99, with prediction errors of 0.3 to 0.4 km/h for running and 5 to 11 W for cycling (full text).
Cycling tertiles by power at LT were ≤ 179 W, 180–209 W and ≥ 210 W in males and ≤ 127 W, 130–160 W and ≥ 162 W in females; FULTP did not differ across tertiles in either sex (full text, Table 3).