Eur J Appl Physiol (2007) 100:645–651

DOI 10.1007/s00421-007-0454-3

ORIGINAL ARTICLE

Cross-validation of the 20- versus 30-s Wingate anaerobic test

C. Matthew Laurent Jr. · Michael C. Meyers · Clay A. Robinson · J. Matt Green

Accepted: 19 March 2007 / Published online: 12 April 2007© Springer-Verlag 2007

Abstract The 30-s Wingate anaerobic test (30-WAT) isthe most widely accepted protocol for measuring anaerobicresponse, despite documented physical side eVects. Abbre-viation of the 30-WAT without loss of data could enhancesubject compliance while maintaining test applicability.The intent of this study was to quantify the validity of the20-s Wingate anaerobic test (20-WAT) versus the tradi-tional 30-WAT. Fifty males (mean § SEM; age = 20.5 §0.3 years; Ht = 1.6 § 0.01 m; Wt = 75.5 § 2.6 kg) wererandomly selected to either a validation (N = 35) or cross-validation group (N = 15) and completed a 20-WAT and30-WAT in double blind, random order on separate days todetermine peak power (PP; W kg¡1), mean power (MP;W kg¡1), and fatigue index (FI; %). Utilizing poweroutputs (relative to body mass) recorded during each sec-ond of both protocols, a non-linear regression equation(Y20WAT+10 = 31.4697 e¡0.5[ln(Xsecond/1174.3961)/2.63692];r2 = 0.97; SEE = 0.56 W kg¡1) successfully predicted(error »10%) the Wnal 10 s of power outputs in the cross-validation population. There were no signiWcant diVer-ences between MP and FI between the 20-WAT thatincluded the predicted 10 s of power outputs (20-WAT+10)

and the 30-WAT. When derived data were subjected toBland–Altman analyses, the majority of plots (93%) fellwithin the limits of agreement (§2SD). Therefore, whencompared to the 30-WAT, the 20-WAT may be considereda valid alternative when used with the predictive non-linearregression equation to derive the Wnal power output values.

Keywords Leg power · Work capacity · Cycle ergometry · Sprint test

Introduction

The most commonly employed protocol for the measure-ment of anaerobic response is the 30-s Wingate anaerobictest (30-WAT). Developed almost 30 years ago, the 30-WAT involves a maximal exertion bout on a cycle ergome-ter to evaluate peak power (PP), mean power (MP), andfatigue index (FI; Bar-Or 1987; Bar-Or et al. 1977). Whenperforming a 30-WAT, a subject typically exhibits a sharprise in power output, reaching peak power within the Wrstfew seconds. Typically, subjects are unable to maintain thisoutput, leading to an exponential decline in power through-out the remaining duration of the test (Bar-Or et al. 1977;Marquardt et al. 1993). During this prolonged period ofmaximal eVort, the accumulation of [H+] and lactate asbyproducts of anaerobic glycolysis results in a drop inblood pH (lactic acidosis). The increased acidity impairsenzyme activity involved in energy metabolism andreduces maximal muscle Wbre recruitment (Allen et al.1992; Davis 1985). In addition, the acute increase of bloodglucose as a substrate for glycolysis during maximal exer-cise can result in hypoglycemia (Vincent et al. 2004).Together these responses may result in unwelcome physicalside eVects, including fatigue, headache, dizziness, and

C. Matthew Laurent Jr. (&) · J. Matt GreenDepartment of Kinesiology, The University of Alabama, Box 83012, Moore Hall, Tuscaloosa, AL, 35487-0312, USAe-mail: cmlaurentjr@bama.ua.edu

M. C. MeyersHuman Performance Research Laboratory, Department of Sports and Exercise Sciences, West Texas A&M University, Canyon, TX 79016, USA

C. A. RobinsonDepartment of Agriculture, West Texas A&M University, Canyon, TX 79016, USA

123

646 Eur J Appl Physiol (2007) 100:645–651

nausea (Jacobs et al. 1982). Subject awareness of these sideeVects during anaerobic testing may result in submaximaleVort, high attrition rates, or unsuccessful subject comple-tion of subsequent sport performance testing (Jacobs et al.1982; Marquardt et al. 1993).

In 1987, Vandewalle et al. reported high correlationcoeYcients between data collected at the 20th and 30th sec-onds of a WAT. Consequently, they proposed to curtail theduration of the WAT in order to diminish the stress of thetest and the involvement of the aerobic metabolism. How-ever, this previous study did not compare the 20-WAT and30-WAT but the data collected at the 20th and 30th secondof the same test. Hence this previous study failed to provethat results of a 20-WAT would have been the same asresults collected during the Wrst 20 s of a 30-WAT becauseof a possible submaximal eVort strategy during a stressful30-WAT. More recently, Marquardt et al. (1993) reportedhigh intraclass correlations (r > 0.95) between results of a20 and 30 Wingate tests in 14 subjects and concluded “thata 20-s Wingate test may be a valid and less strenuous alter-native to the 30-s Wingate test.”

Since its Wrst publication, the 30-WAT has been used inhundreds of studies published in the international literatureand it would be interesting to be able to compare the resultsof a future 20-s anaerobic test with all these experimentaldata. Moreover, it is likely that many athletes have per-formed several traditional 30-WATs and, subsequently,could be reluctant to perform new tests whose results can-not be compared with previous scores. The prediction of30-WAT indices from 20-WAT performances is not indi-vidual but statistical in the study by Vandewalle et al.(1987a). In that study, the statistical prediction of theamount of work performed at the 30th seconds (an equiva-lent of MP) from 20-s data was probably accurate(r = 0.989), however, the regression coeYcient betweenpedal rate at 20th and 30th was lower (r = 0.882). In thepresent study we validate a 20-WAT by comparing the datacollected during 20 and 30 s all-out exercises in 50 sub-jects. Moreover, we propose a method which enables anindividual and accurate prediction of 30-WAT indices fromthe data of a 20-WAT (20-WAT+10).

Methods

Subjects

The participants for this study included 50 male, college-aged students (mean § SEM; age = 20.5 § 0.3 years;Ht = 175.8 § 1.2 cm; Wt = 75.5 § 2.6 kg) with no knowncardiovascular/pulmonary disease, metabolic disorders, ormedical contraindications to exercise as determined by self-response medical history form and interview. Subjects were

randomly selected students who volunteered from univer-sity level physical education classes. Subjects were encour-aged not to participate in any strenuous exercise during the24 h prior to testing, and initially reported to the laboratory4 h postabsorptive. Subjects were fully informed of thenature of the study and provided written, informed consentin accordance with the accepted guidelines of the Institu-tional Review Board of the university and the AmericanCollege of Sports Medicine (1997).

Procedures

All subjects initially reported to the laboratory whereuponeach individual had their height (m) and body mass (kg)determined using calibrated physician scales. Anaerobicresponse was quantiWed utilizing both a 20-WAT and a 30-WAT, performed in a double blind, randomly determinedorder to measure peak power (W kg¡1), mean power(W kg¡1), and fatigue index (%). The maximum powerachieved during the Wrst 5 s of the test is deWned as PP,while MP is deWned as the average power achievedthroughout the trial, and the FI reXects the percent powerdecline during the trial (Bar-Or 1987; Bar-Or et al. 1977).

Both WAT protocols were conducted while pedaling acalibrated Monark model 824E cycle ergometer (MonarkAB, Varberg, Sweden) with integrated laser-based sensorand computer software (Sports Medicine Industries, Inc.,St. Cloud, MN). Prior to both tests, each subject began awarm-up phase consisting of alternating three 30-s intervalsof active rest (pedaling against no resistance at 60 rpm)with three 30-s intervals pedaling against increasing resis-tance of 25, 50, and 75% of test resistance (Vanderfordet al. 2004). Prior research has indicated that this form ofpre-test loading elicits optimal power production duringsupramaximal exercise (Burnley et al. 2005). Test resis-tance was calculated by multiplying the subject’s bodymass (kg) by 0.075. Following completion of the warm-up,the subject continued to pedal at 60 rpm with no resistancefor another 2 min until initiation of the WAT. Following a10-s countdown, the resistance was immediately added andthe subject was verbally encouraged to pedal as fast as pos-sible for 20 or 30 s. Relative power outputs (W kg¡1) weremeasured during each second of the two trials, and PP, MP,and FI were calculated according to accepted procedures(Bar-Or 1987; Bar-Or et al. 1977).

Within each subject, the 20-WAT and 30-WAT werecompleted at the same time of day, no less than 48 h and nomore than 7 days apart in order to ensure optimal recoveryand minimize any confounding anthropometric changes,respectively. In accordance with previously publishedresearch regarding test–retest reliability of the WAT (Bar-Or1987; Granier et al. 1995; Kaczkowski et al. 1982;Vandewalle et al. 1987a), a PP increase ¸8% was considered

123

Eur J Appl Physiol (2007) 100:645–651 647

to be a distorted response, presumably due to a learningeVect. In these instances, the initial protocol was repeated48 h later and used for comparison. In 13 instances, thediVerence between the two protocols still exceeded thestated criteria and, consequently, these subjects were omit-ted from the study. To assess the presence of negative sideeVects following testing, a basic approach was taken. Sub-jects were asked to provide a yes/no answer with respect topresence of nausea, light-headedness, leg fatigue, and/orany other physical side eVects. This approach was advanta-geous in that it was not diYcult to administer and permittedthe evaluation of side eVects immediately proximal to thetest. This was important as side eVects may often be acuteand transient and in such case a written survey completedby subjects after recovering from testing would provide lessvalid information.

Statistical analyses

Data were subjected to multivariate analyses of variance(MANOVAs), where appropriate, utilizing the StatisticalPackage for Social Sciences (SPSS, Inc., Chicago, IL) todetermine a signiWcant main eVect between protocols.When univariate post hoc procedures (AVOVA) indicatedsigniWcant diVerences between relative MP outputs and FI,the sample of 50 subjects were randomly assigned to eithera validation group (N = 35) or cross-validation group(N = 15). Both linear and non-linear regression analyseswere employed to develop an equation in order to deter-mine the best model to predict the Wnal 10 s of relativepower outputs. These values were derived from power out-puts obtained during seconds 11–20 of the 20-WAT rela-tive to those obtained during seconds 21–30 of the 30-WATfrom the validation group. The non-linear formulaemployed to Wt the data was a three parameter, peak, log-normal equation of the form Y = a e¡0.5[(ln (X/X0)/b)2],where a, b, and X0 are the parameters Wtted, e¡0.5 is anexponential function, ln is the command to perform a nor-mal logarithmic procedure, and X is the second for whichthe relative power output is being determined. Followingregression analyses, the predicted Wnal 10 s were combinedwith the observed 20-WAT data (20-WAT+10) recorded inthe cross-validation group to demonstrate the validity andapplicability of the derived equation to a target population.ANOVAs were performed to identify signiWcant diVer-

ences among the PP, MP, and FI of the 20-WAT+10 andthe 30-WAT relative power output. To further assess valid-ity and reliability, the Bland–Altman method of comparisonwas employed to determine the limits of agreement(mean § 2SD) between the two protocols (Altman andBland 1983; Bland and Altman 1986) within the cross-vali-dation group. Statistical signiWcance was determineda priori at the P = 0.05 level. All data are presented asmean § SEM unless otherwise noted.

Results

Peak power, mean power, and fatigue index

The mean relative power outputs recorded during the20-WAT as well as the 30-WAT along with their respectiveP values and percent diVerences are presented in Table 1.As expected, a signiWcant MANOVA was found for theWilks’ Lambda rank variable (F3,80 = 0.723; P < 0.001;n ¡ � = 0.998) between protocols. Subsequent ANOVAsindicated no signiWcant diVerences between relative PP out-puts recorded during the 30-WAT and the 20-WAT; how-ever, signiWcant diVerences were found with regard torelative MP and FI. Mean relative power outputs were sim-ilar throughout the Wrst 20 s of both trials, however, a sub-stantial power decline was observed during the Wnal 10 s ofthe 30-WAT (Fig. 1).

Linear versus non-linear regression prediction model

Results from linear and non-linear regression analysesrevealed that the non-linear model yield similar coeYcientsof determination between models. There was an additionaltwo percent of variation observed was explained whenusing a non-linear versus a linear prediction model(r2 = 0.97 vs. r2 = 0.95, respectively). Additionally, anoticeable discrepancy was revealed in the standard error ofthe estimate between models, with the non-linear modelreporting an error of »10% (SEE = 0.56 W kg¡1) whencompared to an error of »15% (SEE = 0.75 W kg¡1) of thelinear model observed within the measured population.Consequently, the non-linear regression model wasemployed to validate and cross-validate the 20-WAT ver-sus 30-WAT.

Table 1 Mean anaerobic power and capacity for 20-WAT and 30-WAT (N = 50)

Variable 30-WAT Mean § SEM

20-WAT Mean § SEM

P DiVerence (%)

Peak power (W kg¡1) 6.8 § 0.2 6.7 § 0.2 0.54 1.5

Mean power (W kg¡1) 4.5 § 0.1 5.0 § 0.2 0.01* 10.0

Fatigue index (%) 58.3 § 1.7 46.1 § 1.7 <0.01* 20.9* SigniWcant at P < 0.05 level

123

648 Eur J Appl Physiol (2007) 100:645–651

Predicted versus observed MP and FI

The relative power outputs of the 20-WAT+10 compared tothe observed relative power outputs of the 30-WAT for all35 subjects in the validation group (Fig. 2) was modeledand curve Wtted for all individuals in a three parameter,peak, log-normal equation (Y20WAT+10 = 31.4697 e¡0.5 [ln(Xsecond/1174.3961)/2.63692]). The model was signiWcant

and the slope and intercept in the equation were signiW-cantly diVerent from zero (P < 0.001).

The mean relative power outputs recorded during the 20-WAT+10 as well as the 30-WAT, their respective P values,and percent diVerences are presented in Table 2. MANOVAindicated no signiWcant main eVect for the Wilks’ Lambdarank variable (F3,80 = 0.182; P = 0.909; n ¡ � = 0.082)between a 20-WAT+10 and 30-WAT.

Fig. 1 Average relative power outputs of all subjects (N = 50) between 20-WAT and 30-WAT

(N =50)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Second

W·k

g-1

20-WAT 30-WAT

Fig. 2 Average relative power outputs of the validation group (N = 35) between 20-WAT+10 and 30-WAT

(N = 35)

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

1 3 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Seconds

W·kg

-1

30-WAT 20-WAT+10

2 4 5 6 7 8 9

Table 2 Mean anaerobic power and capacity for the 20-WAT+10 and the 30-WAT in the validation group (N = 35)

No values were signiWcantly diVerent at the P < 0.05 level

Variable 30-WAT Mean § SEM 20-WAT+10 Mean § SEM P DiVerence (%)

Peak power (W kg¡1) 6.6 § 0.2 6.6 § 0.2 0.75 0.0

Mean power (W kg¡1) 4.3 § 0.1 4.4 § 0.1 0.71 2.3

Fatigue index (%) 59.4 § 1.7 57.5 § 1.2 0.20 3.3

123

Eur J Appl Physiol (2007) 100:645–651 649

The mean observed and predicted relative power outputsfor the 20-WAT+10 as well as the 30-WAT in the cross-validation group, their respective P values, and percentdiVerences are presented in Table 3. Furthermore, ANO-VAs indicated no signiWcant diVerences between relativePP outputs, relative MP outputs, or FI between the two pro-tocols in the cross-validation group (Fig. 3).

Figures 4, 5 illustrate the limits of agreement (§2SD) asdetermined by the Bland–Altman method of comparison.Predicted values of MP and FI demonstrated acceptableagreement when compared to the actual values of MP and

FI observed during the 30-WAT trials, with 93% of allobservations falling within the §2SD limits of agreement.

Discussion

The 30-WAT has been shown to be highly reliable andapplicable in a wide variety of settings, having test–retestreliability of 0.90–0.97 (Bar-Or 1987; Bar-Or et al. 1977;Kaczkowski et al. 1982; Vandewalle et al. 1987b). The30-WAT has been used to evaluate anaerobic performance

Fig. 3 Average relative power outputs of the cross-validation group (N = 15) between 20-WAT+10 and 30-WAT

(N = 15)

30-WAT 20-WAT+102.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

1 3 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Seconds

W·k

g-1

2 4 5 6 7 8 9

Table 3 Mean anaerobic power and capacity for the 20-WAT+10 and the 30-WAT in the cross-validation group (N = 15)

No values were signiWcantly diVerent at the P < 0.05 level

Variable 30-WAT Mean § SEM 20-WAT+10 Mean § SEM P DiVerence (%)

Peak power (W kg¡1) 6.9 § 0.2 7.1 § 0.3 0.06 2.9

Mean power (W kg¡1) 4.6 § 0.2 4.6 § 0.2 0.58 0.0

Fatigue index (%) 55.4 § 2.4 58.2 § 1.7 0.08 4.9

Fig. 4 Bland–Altman plots of relative mean power outputs illustrating the upper and lower levels of agreement (dashed lines) between 30-WAT and 20-WAT+10

(N = 15)

-2.0-1.8-1.6-1.4-1.2-1.0-0.8-0.6-0.4-0.20.00.20.40.60.81.01.21.41.61.82.0

3.0 3.5 4.0 4.5 5.0 5.5 6.0

Mean Power (W·kg-1)

Ob

serv

ed D

iffe

ren

ce B

etw

een

Pro

toco

ls

123

650 Eur J Appl Physiol (2007) 100:645–651

among children, adolescents, and adults with activity levelsranging from sedentary to athletic (Bar-Or 1987; Groussardet al. 2003; Mastrangelo et al. 2004; Van Someren andPalmer 2003; Vincent et al. 2004). However, the limitationsand applicability of the WAT to a variety of populations,including high-level athletes, has been well documented inother studies (Beneke et al. 2002; Medbø and Tabata,1993). Additionally, detrimental physical side eVects havebeen documented, and may constitute a threat to optimalsubject compliance (Maud and Schultz 1989; Ulmer 1996).Abbreviation of the 30-WAT without loss of data couldpotentially ensure subject compliance while maintainingtest applicability (Marquardt et al. 1993; Smith and Hill1991). Therefore, the purpose of this study was to quantifythe validity of and, subsequently, cross-validate the 20-WAT versus 30-WAT.

As expected, results from this study indicate an abbrevi-ated version of the 30-WAT produced signiWcantly diVer-ent relative MP outputs and FI values. In the study byMarquardt et al. (1993), average power output (PO) fromseconds 15–20 during 20-WAT was apparently correlatedwith PO from seconds 15–20 during 30 WAT instead of POfrom seconds 25–30, which did not enable the computationof FI during 30-WAT. In a study by Vandewalle et al.(1987a), the correlation coeYcient between velocity at 20and 30 s was lower than the correlation coeYcient betweenMP (0.882 vs. 0.989) although these velocities were mea-sured during the same exercise. In the present study, thefatigue indices (FI) corresponding to 20-WAT+10 and 30-WAT were calculated for exercises performed on separatedays. Nevertheless, the diVerences between FI were only3.3 and 4.9% (Tables 2, 3 for the validation and cross vali-dation studies, respectively) although the test–retest corre-lation coeYcients in the literature are low for the fatigueindex. Consequently, it is possible to predict the Wnal 10 s

of relative power outputs and attain similar relative MP out-puts and FI values observed during a traditional 30-WATwith the use of the derived non-linear regression equation.

Comparison of the 20-WAT versus the 30-WAT

As seen in Figs. 1, 2, 3, each subject achieved maximalpower output within the initial 5 s of the anaerobic bouts.This was followed by a gradual decline in relative power, asexpected, throughout the remainder of the test, resulting in aminimal power output during the Wnal 5 s, as observed inearlier studies (Ansley et al. 2004; Bar-Or 1987; Calbet et al.2003; Gastin 2001; Granier et al. 1995; Murphy et al. 1986;Smith and Hill 1991; Vandewalle et al. 1987a). The lack ofsigniWcant diVerence in PP reported in this study indicatesthat similar eVort was demonstrated over both trials.

Despite similar relative power outputs observed duringthe Wrst 20 s of both trials, signiWcant diVerences wereobserved between the overall relative MP outputs and FIvalues between the two protocols, as would be expected.While lactate and acidity were not directly assessed in thecurrent study, it is plausible that these by-products poten-tially associated with fatigue may have diVered between tri-als and consequently aVected performance. However, it isstrongly emphasized that without measures, any conclu-sions as such would be speculative at best.

Validity of the 20-WAT+10 versus the 30-WAT

Following the derivation of the non-linear regression equa-tion, predicted power output and FI values resulted in simi-lar relative MP outputs and FI values between the 20-WAT+10 and the 30-WAT in the validation and cross vali-dation groups (Figs. 1, 2). Both PP and MP values from the20-WAT+10 and 30-WAT met or exceeded the test–retest

Fig. 5 Bland–Altman plots of fatigue index illustrating the upper and lower levels of agreement (dashed lines) between 30-WAT and 20-WAT+10

(N = 15)

-12.0-10.0

-8.0-6.0-4.0-2.00.02.04.06.08.0

10.012.014.016.018.020.022.024.0

45.0 50.0 55.0 60.0 65.0 70.0

Fatigue Index (%)

Ob

serv

ed D

iffe

ren

ce B

etw

een

Pro

toco

ls

123

Eur J Appl Physiol (2007) 100:645–651 651

reliability values previously reported with respect to poweroutputs observed between two separate 30-WATs (Bar-Or1987; Kaczkowski et al. 1982). These Wndings suggest thatadministering a 20-WAT is still advantageous if (a) thederived regression equation is employed to determine rela-tive power outputs of the Wnal 10 s of a WAT to maintainthe validity typically associated with the original protocolor (b) by acknowledging the percent diVerences that existbetween the protocols reported in this study.

Physical response

Detrimental physical responses and subsequent subjectapprehension have been reported to occur both during andafter the 30-WAT, including nausea, dizziness, headaches,and vomiting resulting in less than optimal compliance withthe 30-WAT (Jacobs et al. 1982; Maud and Schultz 1989).Following the 30-WAT in this study physical discomfort wasobserved as three subjects vomited, and approximately 25%subjects reported nausea, light-headedness, headaches, and/or leg fatigue, despite a 4-min cool-down and provision of ahigh-carbohydrate drink upon request. While these eVectscan only be reported subjectively, the only detrimental sideeVect reported following the 20-WAT was leg fatigue. Theminimal side eVects (aside from leg exhaustion) seem toindicate a reduced level of discomfort among subjectsfollowing the 20-WAT, which could increase subject compli-ance as well as the test–retest reliability of the protocol.

Conclusion

The major Wnding from this study is that a 20-WAT can beconsidered a valid alternative to 30-WAT when used con-comitantly with the prediction of the Wnal 10 s utilizing theregression equation derived in this study. The use of thisabbreviated 20-WAT+10 protocol may reduce subject dis-comfort both during and following the test, thereby maximiz-ing subject compliance and enhancing the applicability andrepeatability of the WAT across a variety of populations.

References

Allen DG, Westerblad JA, Lee JA, Lannergren J (1992) Role of exci-tation–contraction coupling in muscle fatigue. Sports Med13:116–126

Altman DG, Bland JM (1983) Measurement in medicine: the analysisof method comparison studies. Statistician 32:307–317

American College of Sports Medicine (1997) Policy statement regard-ing the use of human subjects and informed consent. Med SciSports Exerc 29:5

Ansley L, Robson PJ, Gibson A, Noakes TD (2004) Anticipatory pac-ing strategies during supramaximal exercise lasting longer than30 s. Med Sci Sports Exerc 36:309–314

Bar-Or O (1987) The Wingate anaerobic test: an update on methodol-ogy, reliability and validity. Sports Med 4:381–394

Bar-Or O, Dotan R, Inbar O (1977) A 30 s all-out ergometric test: itsreliability and validity for anaerobic capacity. Isr J Med Sci13:126

Beneke R, Pollmann C, Bleif I, Leithauser RM, Hutler M (2002) Howanaerobic is the Wingate anaerobic test for humans? Eur J ApplPhysiol 87:388–392

Bland JM, Altman DG (1986) Statistical methods for assessing agree-ment between two methods of clinical measurements. Lancet1:307–310

Burnley M, Doust JH, Jones AM (2005) EVects of prior warm-up re-gime on severe-intensity cycling performance. Med Sci SportsExerc 37:838–845

Calbet JA, De Paz JA, Garatachea N, Cabeza de Vaca S, Chavarren J(2003) Anaerobic energy provision does not limit Wingate exer-cise performance in endurance-trained cyclists. J Appl Physiol94:668–676

Davis JA (1985) Anaerobic threshold: review of the concept and direc-tions for future research. Med Sci Sports Exerc 17:6–21

Gastin PB (2001) Energy system interaction and relative contributionduring maximal exercise. Sports Med 31:725–741

Granier P, Mercier B, Mercier J, Anselme F, Prefaut C (1995) Aerobicand anaerobic contribution to Wingate test performance in sprintand middle-distance runners. Eur J Appl Physiol 70:58–65

Groussard C, Machefer G, Rannou F (2003) Physical Wtness and plas-ma non-enzymatic antioxidant status at rest and after a Wingatetest. Can J Appl Physiol 28:79–92

Jacobs I, Bar-Or O, Karlsson J, Dotan R, Tesch P, Kaiser P, Inbar O(1982) Changes in muscle metabolites in females with 30-sexhaustive exercise. Med Sci Sports Exerc 14:457–460

Kaczkowksi W, Montgomery DL, Taylor AW, Klissouras V (1982) Therelationship between muscle Wber composition and maximal anaer-obic power and capacity. J Sports Med Phys Fitness 22:407–413

Marquardt JA, Bacharach DA, Kelly JM (1993) Comparison of poweroutputs generated during 20 and 30 s Wingate tests. Res Q ExercSport 64:A33–A34

Mastrangelo MA, Chaloupka EC, Kang J, Lacke CJ, Angelucci JA,Martz WP, Biren GB (2004) Predicting anaerobic capacity in 11–13 year-old boys. J Strength Cond Res 18:72–76

Maud PJ, Schultz BB (1989) Norms for the Wingate anaerobic test withcomparison to another similar test. Res Q Exerc Sport 60:144–151

Medbø JI, Tabata I (1993) Anaerobic energy release in working muscleduring 30 s to 3 min of exhausting bicycling. J Appl Physiol75:1654–1660

Murphy MM, Patton JF, Frederick FA (1986) Comparative anaerobicpower of men and women. Aviat Space Environ Med 57:636–641

Smith JC, Hill DW (1991) Contribution of energy systems during aWingate power test. Br J Sports Med 25:196–199

Ulmer HV (1996) Concept of an extracellular regulation of muscularmetabolic rate during heavy exercise in humans by psychophysi-ological feedback. Experientia 52:416–420

Van Someren KA, Palmer GS (2003) Prediction of 200-m sprint kaya-king performance. Can J Appl Physiol 28:505–517

Vanderford ML, Meyers MC, Skelly WA, Stewart CC, Hamilton KL(2004) Physiological and sport-speciWc skill response of Olympicyouth soccer athletes. J Strength Cond Res 18:334–342

Vandewalle H, Heller J, Pérès G, Raveneau S, Monod H (1987a) Etudecomparative entre le Wingate test et un test force-vitesse sur ego-cycle. Sci Sports 2:279–284

Vandewalle H, Pérès G, Monod H (1987b) Standard anaerobic exer-cise tests. Sports Med 4:268–289

Vincent S, Berthon P, Zouhal H, Moussa E, Catheline M, Betue-FerrerD, Gratas-Delamarche A (2004) Plasma glucose, insulin and cat-echolamine responses to a Wingate test in physically active wom-en and men. Eur J Appl Physiol 91:15–21

123