In 580 competitive cyclists the best power-based VO2max equation still had 95% limits of agreement of about ±12 mL/kg/min in men; within a trained group power explains under half of the variance
Key Findings
The best equation (FRIEND, from ramp peak power) had r = 0.684, a standard error of 4.0 mL/kg/min and limits of agreement −15.52 to 8.32 in 496 men; in 84 women r = 0.897, limits −16.76 to 14.56.
580 competitive cyclists (496 men, 84 women). Measured VO2max 63.43 ± 5.49 mL/kg/min in men and 54.82 ± 7.02 in women; peak power 5.90 ± 0.56 and 5.19 ± 0.79 W/kg.
FRIEND equation (from ramp peak power and body mass), men: r = 0.684, standard error of estimate 4.0 mL/kg/min, mean error 3.6, limits of agreement −15.52 to 8.32. Women: r = 0.897, standard error 3.1, mean error 1.1, limits −16.76 to 14.56.
The Storer equation, built on sedentary adults, was off by 10.1 mL/kg/min on average in men.
Practical Application
Even with a laboratory ramp peak power and the best-performing equation, an individual trained cyclist's estimate carries about ±12 mL/kg/min: treat every power-derived VO2max as a rough range.
Among athletes of similar level, power separates VO2max poorly (r about 0.68 in men) — two riders with the same W/kg can differ widely, because efficiency and anaerobic contribution differ.
Equations built on untrained people must not be used for trained cyclists.
Limitations
Single laboratory; ramp protocol not standardised to the source equations; small female sample.
The paper's discussion says the equations "underestimate" while its own table shows predicted means above measured in men; the table values are used here. Full text was read.
A classic group-level relationship (Hawley & Noakes 1992, DOI 10.1007/BF01466278, 100 trained cyclists, r = 0.97 between peak power and VO2max) comes from a mixed-sex, wide-range sample and gives no individual error.