REF-877 · Wright et al., 2017 · International Journal of Sports Medicine, Vol. · Paper · PubMed

The Reliability and Validity of the 3-min All-out Cycling Critical Power Test

Blimited evidenceMeasured on a mixed groupWritten before October 2026, being re-checked

Earlier work proposed estimating critical power (CP) from a single 3-min all-out cycling bout.

Here 12 participants each did a ramp test, three fixed-power trials from which CP and W′ were derived (1/time model), and four 3-min all-out tests, two with cadence held constant (isokinetic) and two against a fixed resistance (linear), yielding end power (EP) and work above end power (WEP).

Isokinetic EP did not differ from CP, linear EP did, and WEP differed from W′ under both modes. EP was very repeatable in both modes, WEP less so. The authors' verdict: the isokinetic test estimates CP reliably and validly, the linear test reliably but not validly, and W′ should not be estimated from the 3-min all-out test.

Participants

12 participants (abstract: "12 participants"); MEDLINE indexing lists male adults. Training status is not given in the abstract.

Protocol

Each participant completed 8 tests:

1

A ramp test.

2-4. CP and W′ from three fixed-power tests, fitted with the 1/time mathematical model.

5-8. EP and WEP from four 3-min all-out tests: 2 isokinetic (constant cadence) and 2 linear (fixed resistance).

Key Results

ComparisonResult
EP-isokinetic vs CPno significant difference (P=0.377)
EP-linear vs CPsignificantly different (P=0.004)
WEP-isokinetic vs W′significantly different (P<0.001)
WEP-linear vs W′significantly different (P<0.001)

Coefficient of variation between repeat tests: EP-isokinetic 1.93%, EP-linear 1.17%, WEP-isokinetic 8.44%, WEP-linear 5.39%.

Practical Application

In 12 participants who each did two isokinetic and two linear 3-min all-out tests plus a conventional multi-trial CP test, end power matched CP only in the constant-cadence (isokinetic) mode (P=0.377), differed from it in the fixed-resistance (linear) mode (P=0.004), and the test's work-above-end-power never matched W′ (P<0.001). To "can I use a 3-min all-out test instead of several CP trials?": for CP, yes, if the ergometer holds the cadence constant; the number is repeatable in either mode, but a fixed-resistance version gives a different value, and neither mode gives a usable W′.
End power is repeatable (coefficient of variation 1.17-1.93%), so the same test in the same mode can track change over time even when its agreement with CP is not established.
Inference, not tested here: a smart trainer's "slope" or "level" mode is neither of the two lab modes tested; treat a home 3-min all-out result as its own number until it has been checked against multi-trial CP.
Not from this paper; the base's own practice: SYNTH-53 names the self-paced multi-trial approach (REF-026) as the default for CP and W′.

Limitations

n=12, participants not described in the abstract beyond the MEDLINE indexing.
Validity judged by absence of a significant mean difference, not by individual limits of agreement.
Two lab ergometer modes only; road, smart-trainer and self-selected-gear versions were not tested.
Written from the abstract and record metadata only (full text not read).
How this entry was checked and changed
Why this level

test validation study, n=12