REF-809 · Jones & Vanhatalo, 2017 · Sports Medicine, Vol. · Paper · PubMed · Free full text

The 'Critical Power' Concept: Applications to Sports Performance with a Focus on Intermittent High-Intensity Exercise

B+moderate evidenceMeasured on a mixed groupWritten before October 2026, being re-checked

The power-duration relationship levels off at critical power (CP), which separates outputs at which muscle phosphocreatine, blood lactate and oxygen uptake stabilise from outputs at which they change continuously until exercise intolerance. The work that can be done above CP (W′) is constant but may be used at different rates depending on how far above CP the athlete works.

The review extends the model from constant-power exercise to intermittent exercise: if W′ is spent during work above CP and rebuilt during recovery below CP, intermittent performance depends on four factors, the intensity and duration of the work intervals and of the recovery intervals.

W′ use may be assumed linear, but its reconstitution appears curvilinear with highly variable kinetics between individuals, which led to a W′-balance model; field trials of stochastic exercise indicate it can accurately predict when W′ tends to zero and exhaustion is imminent.

Participants

No participants: narrative review. The abstract does not say how many studies were reviewed.

Protocol

Conceptual review of the two-parameter CP model and its extension to intermittent high-intensity exercise (W′-balance model), with examples from endurance and team sports. No systematic search is described in the abstract.

Key Results

CP separates sustainable outputs (stable phosphocreatine, lactate and oxygen uptake) from outputs where these change continuously until intolerance.
W′, the work that can be done above CP, is constant but may be used at different rates depending on the proximity of the output to CP.
Intermittent performance depends on four factors: the intensity and duration of the work intervals and the intensity and duration of the recovery intervals.
W′ use may be assumed linear; W′ reconstitution may be curvilinear, with kinetics that are highly variable between individuals.
A W′-balance model has been developed; field trials of stochastic exercise indicate it can accurately predict when W′ tends to zero and exhaustion is imminent.
The authors see applications in real-time monitoring of fatigue in endurance and team sports, informing tactics and pacing.

Practical Application

"How do my intervals and my surges in a race drain me?" — this narrative review from the critical-power group explains that work above critical power spends a fixed reserve (W′) that is rebuilt only below CP, so an interval session or a variable-pace race is decided by four things: how hard and how long the work bouts are, and how easy and how long the recoveries are; W′ may be spent linearly but comes back curvilinearly, at rates that vary widely between athletes.
When designing intervals above CP or critical speed, set the recovery as deliberately as the work: its intensity and length decide how much W′ is back for the next rep.
Do not assume two athletes recover W′ at the same rate; individual reconstitution kinetics were highly variable.
Inference, not tested here: in a middle-distance race, each surge above critical speed spends W′ that cannot be rebuilt while the pace stays above it, so a runner who covers every surge may have nothing left for the finish.
Not from this paper; the base's own practice: measured W′ recovery time constants are in REF-131; CP testing is in REF-044 and REF-026.

Limitations

Narrative review; the abstract contains no numbers, so no W′ values, recovery durations or prediction errors can be quoted from this entry.
The W′-balance model's accuracy is stated from field trials that the abstract does not describe (sample, sport, error).
Written from the abstract and record metadata only (the open-access full text was not read).
How this entry was checked and changed
Why this level

authoritative narrative review from the leading critical-power group; not systematic

Content correction 2026-09-27
2026-10-01: independent check against the abstract — wording brought back to what the abstract states.
Full text (checked 2026-10-12)
CP is typically derived from three to five constant-power tests to exhaustion lasting ~2 to ~15 min (or a 3-min all-out test) and may occur at 70–90% of V̇O₂max depending on training status (full text).
In 12 elite male marathoners, critical speed estimated from personal bests between 1500 m and 15 km averaged 5.84 m/s with D′ 328 m, and the marathon was run at 96 ± 2% of critical speed (full text, Table 1).
After exhaustive severe exercise, lowering power to 110% CP allowed only ~30 s more exercise, 90% CP allowed 20 min in two of six subjects (~10 min in the others), and 80% LT allowed all subjects a further 20 min (full text).
W′ (~22 kJ unfatigued) was 8, 14 and 19 kJ after 2, 6 and 15 min of recovery, a curvilinear recovery with a half-time of ~234 s, faster than blood lactate clearance (~1366 s) (full text).
In the W′-balance model the reconstitution time constant is τW′ = 546·e(−0.01·DCP) + 316, giving ~377 s for recovery at 20 W, ~452 s in the moderate domain and ~580 s in the heavy domain, with no net recharge when recovery stays above CP (full text).
With passive recoveries of 18, 30 or 48 s between fixed work bouts, tolerable duration was ~304, ~516 and ~847 s; across work–rest formats of 20–5 s to 60–30 s the time constant ranged from ~212 s (20–20 s) to ~403 s (60–30 s), with model error ~1.6 kJ (within ~10% of W′) (full text).
The field validation used power-meter data from eight trained triathletes: W′BAL at reported exhaustion was 0.5 ± 1.3 kJ versus a minimum of 3.6 ± 2.0 kJ when the task was completed (full text).