Journal Logo

APPLIED SCIENCES: Physical Fitness and Performance

A 3-min All-out Cycling Test Is Sensitive to a Change in Critical Power


Author Information
Medicine & Science in Sports & Exercise: September 2008 - Volume 40 - Issue 9 - p 1693-1699
doi: 10.1249/MSS.0b013e318177871a
  • Free


Recent evidence indicates that the boundary between the heavy and the severe exercise intensity domains can be estimated from a single 3-min bout of all-out cycling (3,20). This boundary, which can be estimated using the similar "physiological landmarks" of the maximal lactate steady state (13,19) and critical power (CP) (10,17), separates work rates for which an elevated but steady state blood (lactate) and V˙O2 can be attained from those work rates that induce a continuous increase in blood (lactate) and the projection of V˙O2 toward peak oxygen uptake (V˙O2peak) (4,17,18). The rationale for the measurement of CP in a single bout of all-out cycling stems from the hyperbolic power-duration relationship, which defines a finite and rate-independent capacity for work (curvature constant, W′) above the power asymptote (CP). During all-out cycling, this limited work capacity is entirely used, so that after ∼2-2.5 min the W′ has been essentially reduced to zero, and consequently the power output measured over the final 30 s of the test (end power, EP) equals CP and the work done above EP (WEP) represents the W′ (20). A single-visit protocol to estimate CP is of significant practical value when intensity-specific normalization of work rate or prediction of time to exhaustion is required for research or diagnostic purposes. We have previously demonstrated that the EP parameter derived from a single-visit 3-min all-out test provides a reliable and robust estimate of CP (3,20,21). The validity could be further tested by examining the sensitivity of the 3-min all-out protocol to detect a change in conventionally measured CP as a result of a training intervention.

The sensitivity of the conventionally estimated CP to detect training-induced changes in the upper limit of the heavy domain has been well documented (7,11,18). An investigation by Gaesser and Wilson (7) applied two types of training interventions, with the first group performing submaximal continuous training (40 min at 50% V˙O2max, 3 d·wk−1) and the second group performing high-intensity interval training (10 × 2 min at V˙O2max, 3 d·wk−1). After 6 wk, both groups showed significant increases in CP by 13% and 15%, respectively, with no change in the W′. Similarly, Poole et al. (18) reported no change in W′ and a 10% increase in CP after 7 wk of high-intensity interval training (10 × 2 min at 105% V˙O2max, 3 d·wk−1). In another study, 8 wk of continuous endurance training (30-40 min at CP, 3 d·wk−1) resulted in a 31% increase in CP (11). Although the W′ appeared to decline after training (−26%) in the same study, this effect did not attain statistical significance (11). Considered together, these findings indicate that high-intensity interval training can increase the power-asymptote of the power-duration curve (CP) while inducing no systematic change in its curvature (W′; 7,18).

If the power-duration parameters established in the novel all-out test are to respond to high-intensity interval training in the same manner as the CP and W′ (7,18), then it would be expected that the 3-min test power profile would shift upward resulting in greater total work done posttraining, but due to the augmented EP (analogous to CP), there would be no systematic change in the WEP (analogous to W′). Therefore, the aim of the present investigation was to test the hypothesis that the change in conventionally estimated CP (ΔCP) after an intervention of high-intensity interval training would not be different from the change evidenced by 3-min all-out test end power (ΔEP). Additionally, it was hypothesized that the same training intervention would have no significant effect on the "capacity for work above CP" as estimated from the all-out (WEP) or conventional (W′) test protocols.



Ten subjects volunteered to take part in the study including two females. The testing and training procedures were approved by the Ethics Committee for Research Procedures at Aberystwyth University in accordance with the Declaration of Helsinki. Testing and training procedures were fully explained before obtaining written consent from each participant. All subjects were habitually active and accustomed to high-intensity exercise and included runners, cyclists, and those involved in general fitness training. The recruitment of the subjects took place in mid-December, but commencement of the study (mid-January) and training intervention (late January/early February) ensured that the subjects had not been involved in structured training in 4-6 wk before the investigation. Five subjects had been involved in previous studies using the 3-min all-out test (3,20,21). Subjects were instructed to be adequately hydrated and not to consume alcohol for 24 h and food or caffeine for 3 h before each testing session. One female subject withdrew from the study due to an injury unrelated to the project; the data presented are therefore from the nine subjects who completed the full testing and training protocol. The characteristics of the nine subjects were the following: age = 29 ± 6 yr, height = 1.77 ± 0.08 m, and body mass = 74.1 ± 11.9 kg.

Experimental design.

The pretraining testing protocol required six visits to the laboratory. Tests were separated by a minimum of 24 h rest and testing was completed within a 14-d period. Subjects first performed a ramp incremental test for the assessment of V˙O2peak and the gas exchange threshold (GET, 2). During the second visit, subjects performed a 3-min all-out test, which served as a familiarization trial and was not included in the data analyses. The following four tests were performed in a random order, including one 3-min all-out test for the measurement of EP and WEP and three constant work rate predicting trials to exhaustion, which were used to determine the CP and W′. After the completion of the 4-wk interval training program, subjects performed the same tests as described for pretraining, excluding the 3-min familiarization trial. Work rates for the CP trials and the resistance setting for the 3-min all-out test were readjusted for the posttraining testing according to the posttraining ramp test results.

Determination of peak oxygen uptake and GET.

All exercise testing was conducted using an electronically braked cycle ergometer (Lode Excalibur Sport, Groningen, The Netherlands). The ergometer seat and handlebars were adjusted for comfort with cyclists' own pedals fitted if required and settings replicated for subsequent tests. The ramp protocol consisted of 3 min of unloaded baseline pedaling, followed by a ramp increase in power output of 30 W·min−1 until volitional exhaustion. Subjects were instructed to maintain their preferred cadence (70-90 rpm) for as long as possible. The test was terminated when the pedal rate fell more than 10 rpm below the chosen cadence for more than 10 s despite strong verbal encouragement. During this and all subsequent tests, pulmonary gas exchange was measured breath-by-breath. The subjects wore a nose clip and breathed through a low dead space (90 mL), low resistance (0.75 mm Hg·L−1·s·−1 at 15 L·s−1) mouthpiece and impeller turbine assembly (Jaeger Triple V). The inspired and expired gas volume and gas concentration signals were continuously sampled at 100 Hz, the latter using paramagnetic (O2) and infrared (CO2) analyzers (Jaeger Oxycon Pro, Hoechberg, Germany) via a capillary line connected to the mouthpiece. These analyzers were calibrated before each test with gases of known concentration, and the turbine volume transducer was calibrated using a 3-L syringe (Hans Rudolph, MO). The volume and concentration signals were time aligned by accounting for the delay in capillary gas transit and analyzer rise time relative to the volume signal. Oxygen uptake, carbon dioxide output, and minute ventilation were calculated using standard formulae (1) and displayed breath-by-breath. The V˙O2peak was determined as the highest average V˙O2 over a 30-s period. The data were reduced to 10-s averages for the estimation of GET using the V-slope method (2).

Three-min all-out tests.

Subjects first performed a 5-min warm-up at ∼90% GET, followed by 5 min of rest. The test began with 3 min of unloaded baseline pedaling followed by a 3-min all-out effort. Subjects were asked to accelerate to 110-120 rpm over the last 5 s of the baseline period. The resistance on the pedals during the 3-min all-out effort was set using the linear mode of the ergometer so that the subject would attain the power output halfway between the GET and the V˙O2peak (50% Δ) on reaching their preferred cadence (linear factor = power/cadence2). Strong verbal encouragement was provided throughout the test, although the subjects were not informed of the elapsed time in an attempt to prevent pacing. To ensure an all-out effort, subjects were instructed to attain their peak power output as quickly as possible from the start of the test and to maintain the cadence as high as possible at all times throughout the 3 min. To ensure test validity, all subjects were required to undergo a full 3-min all-out familiarization trial before the actual 3-min test data collection. It is noteworthy that while a total of 133 of these all-out tests have been carried out in our laboratory during our recent investigations (3,20,21, present study), in remarkably few instances (10 cases), the test data have been deemed "invalid" due to voluntary pacing of effort during the test, and only once a subject terminated the test prematurely due to intolerable discomfort.

CP protocol.

The CP and W′ were estimated from three predicting trials. The work rates for these trials were equivalent to 70% and 80% Δ and 100% V˙O2peak, estimated to yield times to exhaustion from ∼3 to 9 min. Each trial was preceded by a 5-min warm-up at ∼90% GET and a 5-min rest period, followed by 3 min of unloaded baseline pedaling. Subjects were instructed to maintain their preferred cadence for as long as possible during a test, which was terminated when cadence fell more than 10 rpm below the preferred cadence for more than 10 s. Strong verbal encouragement was provided throughout the test, and time to exhaustion recorded to the nearest second. Subjects were not informed of the work rates or their performance in any of the tests until the entire study had been completed. Linear regression was used to provide two sets of CP and W′ estimates from the results of these trials, using the work-time (W = CPt + W′) and the 1/time (P = W′(1/t) + CP) models. Two sets of parameter estimates were calculated and compared to ensure the accuracy of the method. It is acknowledged that the accuracy of the parameter estimates is improved by administering further predicting trials. However, due to the extremely demanding nature of the exercise testing and the frequent laboratory visits required of the participants over the training and testing periods, the number of trials was limited to three. In the present study, the good linear fit of data and the agreement in parameter estimates between the different model fits justify this approach (see Results section).

Training intervention.

The training intervention consisted of supervised sessions on the Lode Excalibur Sport ergometer (enabling external control of power output independent of pedal rate) undertaken three times a week for a 4-wk period. In two of the weekly sessions, subjects completed six 5-min intervals at 105% of EP separated by 2.5 min of active recovery, and in one weekly session, subjects completed 10 repetitions of 2-min intervals separated by 2 min of active recovery. The work rate for the 2-min intervals was calculated so that 50% of WEP would be expended during the first 2-min effort (P = 50%WEP/120 s + EP). If the heart rate was more than 10 bpm below maximum during the last effort of any session, the work rate for the subsequent session was increased after consultation with the subject. In addition, the subjects were permitted to continue their habitual activities outside the supervised sessions, provided that the subjects were rested for at least 12 h before the supervised training. This amounted to 135 ± 40 min·wk−1 of endurance type exercise (running, cycling, swimming) and 90 ± 11 min·wk−1 of resistance exercise (martial arts, weight lifting, yoga). We permitted these activities in an attempt to improve training compliance and because we were interested in the degree rather than the mechanism of change in CP after the intervention.

Data analyses.

As described previously (3,20), the all-out test end power (EP) was calculated as the average power output over the final 30 s of the test and the WEP as the power-time integral above end power. Peak V˙O2 during the all-out test was calculated as the highest 30 s average achieved during the test (see above). Differences between pretraining and posttraining conditions were assessed using paired-samples t-tests. Relationships were assessed using Pearson product moment correlation coefficients. Statistical significance was accepted at P < 0.05 level and results are reported as mean ± SD unless otherwise stated.


After the training intervention, subjects achieved a higher V˙O2peak (by 0.34 L·min−1 or 10%) and had a higher GET (by 0.36 L·min−1 or 19%) in the ramp incremental test (Table 1). The maximum power attained in the ramp test increased by 7%, and the power output associated with the GET increased by 26% (Table 1). No change in body mass was observed after the 4-wk intervention (74.1 ± 11.9 kg pretraining; 73.6 ± 12.2 kg posttraining).

Parameters of the ramp incremental test, the 3-min all-out test and the work-time relationship before and after 4 wk of interval training.

Figure 1 shows the 3-min test power profiles of a subject before and after the training intervention. The 3-min all-out test EP increased in all subjects after training (mean increase ∼10%; t8 = 6.26, P < 0.001). There was no change in the WEP from pretraining to posttraining (t8 = 1.89, P = 0.10), but the total work done over 3 min was greater after training and the peak V˙O2 measured during the all-out test increased by ∼7% (Table 1). The V˙O2peak measured during the 3-min all-out test was not significantly different from that measured during the ramp test both before (t8 = 1.36, P = 0.21) and after (t8 = 1.88, P = 0.097) training. As shown in Figure 2, V˙O2 attained V˙O2peak within ∼90-120 s both before and after training and remained at these levels until the termination of the test, as noted previously (3,21). The increase in EP was not related to the increase in the power output at GET (r = 0.37, P = 0.32) nor to the increase in maximum power attained in the ramp incremental test (r = 0.29, P = 0.45). The ergometer resistance setting for the all-out phase was recalculated after training taking into consideration the increase in 50% Δ power output, thus the end test power output was generated at the same cadence pretraining and posttraining (81 ± 6 rpm for both pretraining and posttraining; t8 = 0.49, P = 0.64). The peak power output attained during the all-out test increased by 8% (t8 = 4.06, P = 0.004; Table 1), with no change in peak cadence after training (144 ± 10 pretraining and 143 ± 6 rpm posttraining).

The 3-min all-out test power profile of a subject before and after training (panel A) and the linear transformations of the constant work rate predicting trial data using the work-time model (panel B) and the 1/time model (panel C) for the same subject. Closed circles represent pretraining data; open circles represent posttraining data.
The oxygen uptake responses to the 3-min all-out test before(closed circles) and after (open circles) 4 wk of training in a representative subject. The solid horizontal line represents the V˙O2peak measured during ramp exercise before training, and the dashed line represents the V˙O2peak measured by the ramp test after training. Notice the similarity in V˙O2peak achieved during ramp and all-out exercise before and after training, as shown in Table 1. Also note that the V˙O2 remained at V˙O2peak for the last 60-90 s of the test despite the continuing fall in power output.

The power-duration parameter estimates correlated well between the work-time and the 1/time models (CP: r = 0.999 pretraining and r = 0.999 posttraining, and W′: r = 0.991 pretraining and r = 0.982 posttraining) indicating that there was no systematic error in the predicting trial data. The derivation of the parameter estimates using the work-time and 1/time models is shown for a representative individual in Figure 1 (panels B and C). The data from the work-time model were used for subsequent analyses due to better fit of linear data (R2 values ranging between 0.998-1.000 pretraining and 0.999-1.000 posttraining for work-time model and between 0.977-1.000 pretraining and 0.974-0.999 posttraining for 1/time model).

CP increased in all subjects after training (t8 = 7.47, P < 0.001; Table 1). The training intervention had no statistically significant effect on the W′ (t8 = 2.03, P = 0.08; Table 1), although this parameter was reduced in eight out of nine subjects after training. The ΔCP was inversely related to the ΔW′ (r = −0.75, P = 0.02). The all-out test EP was not different from CP (t8 = 1.02, P = 0.34 pretraining; t8 = 1.28, P = 0.24 posttraining), and the EP and CP estimates were highly correlated (r = 0.96, P < 0.001 pretraining; r = 0.95, P < 0.001 posttraining; Fig. 3). The typical error between EP and CP was calculated to be 10.6 W (4.6%) before training and 10.8 W (4.3%) after training (cf. (20)). The ΔEP was correlated with (r = 0.77, P = 0.016) and not different from the ΔCP (t8 = 0.60, P = 0.57). Similarly, the WEP was not different from the W′ (t8 = 0.58, P = 0.58 pretraining; t8 = 1.67, P = 0.13 posttraining), and the two parameters were correlated before training (r = 0.82, P = 0.007) but not after training (r = 0.63, P = 0.07). The ΔWEP was different from the ΔW′ (t8 = 2.99, P = 0.02).

The relationship between the CP and the 3-min all-out test EP before (A) and after training (B).


The principal finding of the present investigation, in agreement with the first hypothesis, was that the increase in the 3-min all-out test EP closely reflected the increase in the independently estimated CP after 4 wk of high-intensity interval training. The demonstrated sensitivity of the EP parameter in detecting a change in CP over time provides additional evidence that the 3-min test EP represents CP. Also in agreement with the second hypothesis, the training intervention had no significant effect on the finite capacity for work above CP as estimated from the all-out (WEP) or conventional (W′) protocols.

Monod and Scherrer (14) originally contended that the CP was a parameter that could not be directly measured in a single exercise test. There is now, however, a substantial body of evidence showing that the EP parameter measured in a single bout of all-out exercise provides a valid and reliable estimate of CP that demarcates the boundary between the heavy and the severe intensity domains (3,20,21). Specifically, we have recently presented evidence that a 3-min all-out test against a fixed resistance can be used to determine the CP in cycling (20), and that exercise above and below the estimate of CP given by such a test results in physiological response profiles consistent with a delayed steady state (below) and a nonsteady state (above) (3). Thus, the power-duration relationship (8-10, 14-16) appears to be generalizable to all-out exercise, at least for cycling. The robustness of the 3-min all-out test to manipulation of cadence and changes in the power profile has also been demonstrated (21). The present study was designed to extend these findings by imposing a training-induced change in the conventionally estimated CP to determine whether the end-test power in the all-out test would respond similarly.

This study has shown that the 3-min all-out test is capable of reflecting changes in both V˙O2peak and CP. The CP was increased when calculated from repeated bouts of exhaustive exercise, as shown previously (7,18), and the V˙O2peak measured during ramp exercise was also substantially improved. These improvements were mirrored by increases in EP and V˙O2peak measured in the 3-min all-out test (by ∼10% and ∼8%, respectively). Our previous work showed that V˙O2peak (3,21) or V˙O2 values close to V˙O2peak (20) are achieved during the 3-min all-out test. The present results extend these findings by demonstrating that the 3-min all-out test is sensitive to increases in V˙O2peak induced by training (Table 1, Fig. 2). In addition to supporting our previous findings that the EP and the CP were not different (20), the increase in the 3-min test EP was strongly correlated with the increase in CP. Consequently, the EP remained similar to the CP after training, suggesting that a 3-min all-out test is also sensitive to training-induced changes in CP. To our knowledge, this is the first study to show that all-out exercise testing is sensitive to changes in both maximal and submaximal markers of aerobic function, although this should not be considered surprising given the theoretical rationale for using all-out exercise to establish CP (3,20) and the fact that the power profile of the test means that the necessary "stimulus" for the attainment of V˙O2peak (severe intensity exercise) is present for the duration of the test.

A further important observation of this study was that the work done above EP during the 3-min all-out test remained unchanged after 4 wk of training. As expected, subjects were able to sustain a higher power output throughout the test thus resulting in a greater total work done, but due to the elevated EP, there was no change in the WEP. Consistent with previous findings, the training intervention had no statistically significant effect on the conventionally estimated W′ (7,18), unlike sprint interval training (12). However, in the present study, the W′ was reduced after interval training in eight of the nine subjects (by ∼1.7 kJ, P = 0.08). Such a trend has been observed previously (11), but the WEP did not follow this trend (Table 1) and consequently did not correlate with W′ after training (although it was not significantly different). Thus, the WEP seems to be insensitive to the enhancement of aerobic function, whereas training-induced changes in CP may result in no change (18) or tend to reduce W′ (11). This may also explain the finding of a significant difference between the change in W′ (−1.7 kJ) and the change in WEP (+0.5 kJ) in spite of neither parameter showing a statistically significant change. However, there is, at present, a great deal of uncertainty regarding the physiological determinants of W′ (4-6). The most basic and mathematically defensible definition of this parameter (that is, the amount of work that can be completed above CP) is common to the WEP, and thus these parameters should be equivalent (3,20). More broadly, W′ has been interpreted as the capacity to perform work anaerobically (9,16). However, it has recently been suggested that the precise value of W′ is dependent on the available anaerobic capacity, V˙O2peak, and the severe-intensity V˙O2 kinetics (4). Each of these factors can be changed independently and/or in different proportion, which makes the prediction of the effect of an intervention on the W′ complicated. Thus, the cause of the seemingly different "effects" of training on the W′ and WEP cannot be ascertained from the present results.

It is acknowledged that the conclusions that can be drawn from the present data may appear to be limited by the exclusion of a control group from the experimental design. This was a deliberate decision because our purpose was not to test the efficacy of the training intervention but rather to use the training intervention to stimulate physiological change and to determine whether the two independent tests (traditionally determined CP and the 3-min all-out test) showed similar responses. The participants were therefore acting as self-controls in a repeated-measures design. A possible methodological limitation is that we altered the flywheel resistance used in the 3-min all-out tests from pretraining to posttraining based upon the changes in the GET and V˙O2peak measured in the ramp test (see Methods section), and therefore the subjects pedalled against a higher resistance posttraining. It is important to note that as a result of the strict standardization of the flywheel resistance, the subjects performed the 3-min all-out tests at the same cadence pretraining and posttraining. Had the resistance setting not been altered, posttraining subjects would have performed the test at a higher than standard cadence, which has been shown to result in reduced estimates of the EP and WEP (21).

In summary, it has been shown that after a period of high-intensity interval training, the change in the 3-min all-out test EP was of the same magnitude as the change in the independently estimated CP. This finding has significant practical implications for the demarcation of heavy and severe exercise intensity domains and provides an important addition to the application of the power-duration relationship to all-out exercise. Thus, from the results of the present study, as well as our previous work (3,20,21), we contend that a 3-min all-out cycling test is a reliable, valid, and robust means of establishing the CP in a single exercise test.


1. Beaver WL, Wasserman K, Whipp BJ. On-line computer analysis and breath-by-breath display of exercise function tests. J Appl Physiol. 1973;34:128-32.
2. Beaver WL, Wasserman K, Whipp BJ. A new method for detecting anaerobic threshold by gas exchange. J Appl Physiol. 1986;60:2020-7.
3. Burnley M, Doust JH, Vanhatalo A. A 3-min all-out test to determine peak oxygen uptake and the maximal steady state. Med Sci Sports Exerc. 2006;38(11):1995-2003.
4. Burnley M, Jones AM. Oxygen uptake kinetics as a determinant of sports performance. Eur J Sport Sci. 2007;7:63-79.
5. Ferguson C, Whipp BJ, Cathcart AJ, Rossiter HB, Turner AP, Ward SA. Effects of prior very-heavy intensity exercise on indices of aerobic function and high-intensity exercise tolerance. J Appl Physiol. 2007;103:812-22.
6. Fukuba Y, Miura A, Endo M, Kan A, Yanagawa K, Whipp BJ. The curvature constant parameter of the power-duration curve for varied-power exercise. Med Sci Sports Exerc. 2003;35(8):1413-8.
7. Gaesser GA, Wilson LA. Effects of continuous and interval training on the parameters of the power-endurance time relationship for high-intensity exercise. Int J Sports Med. 1988;9:417-21.
8. Gaesser GA, Carnevale TJ, Garfinkel A, Walter DO, Womack CJ. Estimation of critical power with nonlinear and linear models. Med Sci Sports Exerc. 1995;27(10):1430-8.
9. Hill DW. The critical power concept. A review. Sports Med. 1993;16:237-24.
10. Hill DW, Poole DC, Smith JC. The relationship between power and the time to achieve V˙O2max. Med Sci Sports Exerc. 2002;34(4):709-14.
11. Jenkins DG, Quigley BM. Endurance training enhances critical power. Med Sci Sports Exerc. 1992;24(11):1283-9.
12. Jenkins DG, Quigley BM. The influence of high intensity exercise training on the Wlim-Tlim relationship. Med Sci Sports Exerc. 1993;25(2):275-82.
13. Jones AM, Doust JH. The validity of the lactate minimum test for determination of the maximal lactate steady state. Med Sci Sports Exerc. 1998;30(8):1304-13.
14. Monod H, Scherrer J. The work capacity of a synergic muscular group. Ergonomics. 1965;8:329-38.
15. Moritani T, Nagata A, deVries HA, Muro M. Critical power as a measure of physical work capacity and anaerobic threshold. Ergonomics. 1981;24:339-50.
16. Morton RH. The critical power and related whole-body bioenegetic models. Eur J Appl Physiol. 2006;96:339-54.
17. Poole DC, Ward SA, Gardner GW, Whipp BJ. Metabolic and respiratory profile of the upper limit for prolonged exercise in man. Ergonomics. 1988;31:1265-79.
18. Poole DC, Ward SA, Whipp BJ. The effects of training on the metabolic and respiratory profile of high-intensity cycle ergometer exercise. Eur J Appl Physiol. 1990;59:421-9.
19. Pringle JSM, Jones AM. Maximal lactate steady state, critical power and EMG during cycling. Eur J Appl Physiol. 2002;88:214-26.
20. Vanhatalo A, Doust JH, Burnley M. Determination of critical power using a 3-min all-out cycling test. Med Sci Sports Exerc. 2007;39(3):548-55.
21. Vanhatalo A, Doust JH, Burnley M. Robustness of a 3-min all-out cycling test to manipulations of power profile and cadence in humans. Exp Physiol. 2008;93:383-90.


©2008The American College of Sports Medicine