You finish a 20-minute test, multiply the average power by 0.95, and enter the result as your FTP. A week later, a platform fits a critical power model to your power-duration curve and returns a number that is several watts different. Same legs, same power meter, two values that both claim to describe the hardest effort you can hold. The disagreement is not a measurement error to be resolved. It is a signal that the two numbers describe different constructs, built from different definitions and different data.
The useful question is not which number is correct. It is which definition matches the session you are about to prescribe. Answering that requires understanding what each number actually is, why they diverge, and what the comparison studies show about the size and direction of the gap.
What FTP is, and what it is not
Functional Threshold Power is defined as the highest power a rider can maintain in a quasi-steady state without fatiguing for approximately one hour (Allen and Coggan, 2010). In practice, almost nobody tests it with a full 60-minute maximal effort. The standard field protocol is a 20-minute maximal time trial, with FTP estimated as 95% of the mean power from that effort. The 0.95 factor is a fixed correction intended to approximate the gap between 20-minute and 60-minute power for a typical trained rider.
That correction factor is where the first source of divergence lives. It is a population-level approximation applied to an individual. A rider with a large anaerobic contribution can produce a strong 20-minute effort that overstates what they can hold for an hour. A rider with a smaller anaerobic reserve may find the 0.95 factor underestimates their true hour power. FTP carries no term that distinguishes between these two riders.
FTP’s strength is not physiological precision. It is standardization. Training zones, workout files, Training Stress Score, and most analytics platforms are built around it. When you serialize a workout as a fraction of FTP, it travels cleanly between devices and platforms. That compatibility is why FTP remains the default reference for day-to-day training structure, even among coaches who understand its limitations.
What critical power is
Critical power comes from a different tradition. The two-parameter model describes the relationship between power output and time to exhaustion as a hyperbola. In its linearized form, total work done is plotted against time, and the slope of that line is CP while the intercept is W-prime (W′). CP is the rate associated with the maximal aerobic steady state, and W′ is the fixed amount of tolerable work above CP available without recovery (Chorley and Lamb, 2020).
The physiological significance of CP is that it demarcates the boundary between the heavy and severe intensity domains. Below CP, steady-state values for muscle phosphocreatine, blood lactate, and pulmonary oxygen uptake can be attained. Above CP, these variables do not demonstrate steady-state behavior; they change continuously until the limit of tolerance (Jones and Vanhatalo, 2017). This is not a convention or a field estimate. It is a physiological landmark.
W′ is the second parameter, and it is what FTP lacks entirely. It represents the finite work capacity above CP. Ride above CP and W′ drains. Drop below CP and it reconstitutes, though the kinetics of that reconstitution are curvilinear and variable between individuals (Jones and Vanhatalo, 2017). The CP-W′ pair does not just give you a sustainable power; it gives you a model of how much you can spend above that power and for how long.
Why the two numbers diverge
The divergence has three sources: different definitions, different protocols, and the presence or absence of a W′ term.
The definitions differ. FTP is anchored to roughly one-hour quasi-steady-state power. CP is the asymptote of the power-duration relationship, which represents the boundary between heavy and severe intensity. These are related but not identical constructs. Research on the maximal metabolic steady state suggests CP often sits at or slightly above the power most riders can hold for a full hour, while FTP is explicitly anchored to that hour. Depending on individual physiology, CP can read above or below FTP.
The protocols differ. CP is conventionally derived from three to five separate high-intensity tests on different days, with power outputs chosen to produce exhaustion between roughly 2 and 15 minutes (Jones and Vanhatalo, 2017). FTP is derived from a single 20-minute effort with a fixed correction. Change the durations used in the CP fit and you change the estimate. Short efforts anchor W′ and can pull a short-protocol CP estimate upward for riders with large anaerobic capacity.
The W′ term matters most. Two riders can produce the same 20-minute average power and therefore the same FTP, while having materially different CP. A rider with a large W′ will have short efforts inflated by anaerobic capacity, which can pull a short-protocol CP estimate upward even if their genuinely sustainable power is lower. FTP cannot see this distinction because it has no W′ term. The CP-W′ pair is the only one of the two frameworks that surfaces it.
What the comparison studies show
The evidence does not support a universal percentage gap. It supports correlation without interchangeability, with cohort-specific biases and wide limits of agreement.
In a cohort of 17 moderately trained cyclists and triathletes, CP was significantly higher than FTP, with a mean bias of 7 ± 13 W and 95% limits of agreement of −19 to 33 W (Karsten et al., 2021). The correlation was strong (r = 0.969), but the authors concluded that the values generally should not be used interchangeably.
In a separate cohort of highly trained cyclists and triathletes, CP and FTP were also not interchangeable. CP was consistently higher than FTP, with a mean bias of 16 W (McGrath et al., 2021). The authors framed that 16 W as a posited correction factor for that homogenous group, not as a general conversion.
These biases are cohort-specific. They do not establish a universal rule that CP is always higher than FTP, or that the gap is a fixed percentage. Direction and magnitude vary with protocol, cohort, and model choice. What the studies do establish is that the two numbers are correlated but not equivalent, and that treating them as interchangeable introduces error into pacing and prescription.
One further point from the critical power literature: a thesis examining maximal lactate steady state, FTP, and CP concluded that CP was the only marker examined that validly represented the maximal metabolic steady state or the upper boundary of the heavy intensity domain (Wong, 2024). That does not mean FTP is useless. It means FTP is a field estimate of a related but distinct construct, and CP is the physiologically anchored parameter.
A practical division of labor
The programming implication is not to replace FTP with CP or vice versa. It is to use each for what it is good at.
Use FTP for zone-based training load and workout standardization. The ecosystem is built on it. Training Stress Score, normalized power, and most workout files are computed relative to FTP. If you serialize a session as a fraction of FTP, it travels cleanly between your head unit, your indoor trainer, and your analytics platform. Keeping zones on one reference prevents the confusion that comes from mixing FTP-based and CP-based prescriptions in the same plan.
Use CP and W′ for pacing surge-heavy efforts and for diagnosing whether the aerobic ceiling or the anaerobic reserve moved. If you know your W′ is 20 kJ and you spend 15 kJ covering attacks on a climb, you know you have very little left for the summit. That is information FTP cannot give you. CP also separates your aerobic ceiling from your anaerobic capacity, so when you retest you can see which one actually improved. A flat FTP across a training block looks like no progress, but if CP held steady while W′ grew, you did improve, just in a different system.
Record your personal offset between the two numbers. The size and direction of that offset is diagnostic information. A stable offset with a moving CP or W′ tells a different story than a stable offset with everything flat. If FTP stalls, check whether CP or W′ moved before concluding that training has stopped working.
Testing hygiene
A CP fit is only as trustworthy as the maximal efforts feeding it. If you never go truly maximal for 3 to 5 minutes, the model has nothing to anchor W′ with, and your CP estimate becomes unreliable. The same applies to FTP: a 20-minute test that was not truly maximal, or that was performed in a fatigued state, produces a number that will mislead every zone derived from it.
For CP, use at least two maximal efforts of different durations, performed fresh. A common pairing is around 3 minutes and 12 minutes, sometimes with a 5 to 7 minute effort added for a better fit. Do them on separate days or with very long recovery between, because a fatigued second effort corrupts the model. The sources describe ranges rather than a universal standard: two to seven laboratory visits for traditional constant-work-rate testing, or a single 3-minute all-out test that provides valid estimates of CP and W′ from one effort (Chorley and Lamb, 2020).
For FTP, the standard is a single 20-minute maximal effort after a thorough warm-up, on a steady road or a trainer in a stable environment. Take the average power, multiply by 0.95, and that is your FTP. Some riders prefer the 8-minute protocol, taking 90% of a single 8-minute effort, but it is noisier and tends to overestimate. Ramp tests are convenient on smart trainers but read high for riders with strong neuromuscular power.
Whichever you choose, the same warm-up, terrain, and pacing discipline matter. Garbage efforts produce garbage thresholds, and that is true for both models.
What to do when the numbers disagree
First, stop treating the disagreement as an error. It is expected. The two numbers describe different constructs, built from different definitions and different data. The question is not which one is right but which one matches the session.
Second, note the direction and magnitude of your personal offset. If CP reads higher than FTP, that is consistent with the comparison studies, but the size of the gap is yours. If CP reads lower, that is also possible depending on your physiology and the protocol used. Record it and watch how it changes over time.
Third, keep your zones on one reference so that workout files travel cleanly. FTP is the practical choice for most riders because the ecosystem is built on it. Use CP and W′ as a diagnostic layer on top, not as a replacement for the reference that your devices and platforms expect.
Fourth, retest both every 4 to 8 weeks. If FTP stalls, check whether CP or W′ moved. Adjust training emphasis toward whichever system has stopped responding. A rider whose CP is flat but whose W′ is growing is making progress in a different system, and the single FTP number hides that.
FAQ
Is CP always higher than FTP?
No. The comparison studies found CP higher than FTP in their cohorts, with mean biases of 7 W and 16 W, but direction and magnitude vary with protocol, cohort, and model choice. The evidence does not support a universal rule.
Can I convert FTP to CP with a fixed correction factor?
Not for a general population. The 16 W bias reported by McGrath et al. (2021) was cohort-specific and framed as a posited correction for a homogenous group. A fixed conversion factor applied across riders would introduce error rather than reduce it.
Which number should I use for interval prescription?
Use FTP for zone-based training load and workout standardization, because the ecosystem is built on it. Use CP and W′ for pacing surge-heavy efforts and for diagnosing whether the aerobic ceiling or the anaerobic reserve moved. Keep zones on one reference so that files travel cleanly between devices.
How many tests do I need for a valid CP estimate?
At least two maximal efforts of different durations, performed fresh. A common pairing is around 3 minutes and 12 minutes, sometimes with a 5 to 7 minute effort added for a better fit. Traditional constant-work-rate testing uses three to five tests on separate days, but a single 3-minute all-out test can also provide valid estimates (Chorley and Lamb, 2020).
Why does my CP estimate change when I use different test durations?
Because short efforts anchor W′ and can pull a short-protocol CP estimate upward for riders with large anaerobic capacity. The model is only as good as the efforts feeding it. Change the durations and you change the estimate.
Sources
Allen, H. and Coggan, A. (2010). Training and Racing with a Power Meter. VeloPress. Definition of FTP as the highest power a rider can maintain in a quasi-steady state without fatiguing for approximately one hour.
Chorley, A. and Lamb, K.L. (2020). The Application of Critical Power, the Work Capacity above Critical Power (W′), and Its Reconstitution: A Narrative Review of Current Evidence and Implications for Cycling Training Prescription. Sports, 8(9), 123. https://pmc.ncbi.nlm.nih.gov/articles/PMC7552657
Jones, A.M. and Vanhatalo, A. (2017). The ‘Critical Power’ Concept: Applications to Sports Performance with a Focus on Intermittent High-Intensity Exercise. Sports Medicine, 47(Suppl 1), 65–78. https://pmc.ncbi.nlm.nih.gov/articles/PMC5371646
Karsten, B., Petrigna, L., Klose, A., Bianco, A., Townsend, N., and Triska, C. (2021). Relationship Between the Critical Power Test and a 20-min Functional Threshold Power Test in Cycling. Frontiers in Physiology, 11, 613151. https://www.frontiersin.org/journals/physiology/articles/10.3389/fphys.2020.613151/full
McGrath, E., Mahony, N., Fleming, N., Raleigh, C., and Donne, B. (2021). Do Critical and Functional Threshold Powers Equate in Highly-Trained Athletes? International Journal of Exercise Science, 14(4), 45–59. https://pmc.ncbi.nlm.nih.gov/articles/PMC8136559/
Wong, S. (2024). A critical appraisal of methods to determine the Maximal Metabolic Steady State in cycling. PhD thesis, University of Kent. https://doi.org/10.22024/unikent/01.02.104738