There is a relatively flat power-cadence relationship between 60 and 100 rpm in fatigued subjects (Beelen and Sargeant 1991). Does this mean power output measured at 80-100 rpm can be compared with power output at 40 rpm? Burnley (2010) says no, but Marcora and Staiano would argue that this is acceptable.There is published support for Burnley's point. Acceleration at the end of a Wingate test yields a linear torqueangular velocity relationship, which predicts a parabolic power-cadence relationship in these fatigued subjects (MacIntosh et al. 2004). In fatigue, power output would be greater when measured at 75 rpm than when measured at 40 rpm.We would argue that the results of Beelen and Sargeant (1991), which are remarkably similar to the results of Marcora and Staiano (2010a, b), are consistent with those of MacIntosh et al. (2004). The actual power output for the corresponding torque-cadence relationship has been added to Fig. 1, and these demonstrate that there is variation across a broad range of cadences. If you were to test at discreet cadences from 60 to 120 rpm, on separate days, as Beelen and Sargeant (1991) did, you would obtain power outputs distributed around the lines presented. The results shown in Fig. 1 illustrate that this range of cadences will yield similar power outputs: they exist on the plateau of the fatigued parabolic power-cadence relationship. Clearly, if you were to test at 40 rpm, the power drops off considerably.Marcora and Staiano (2010a, b) would have us believe that their subjects with a maximal power output of just over 1,000 W were capable of generating a power output over 700 W at 40 rpm, in the fatigued state. Note that the maximal predicted power in the non-fatigued state at 40 rpm for our subject (maximal power output of 1,248 W) is 740 W. It should also be pointed out that the parabolic power-cadence relationship overestimates power at 40 rpm because the maximum (isometric) torque that can be generated (while seated) is body weight (in Newton) times crank length. For a 175 mm crank and 82 kg subject, the maximum (isometric) torque would be 140 Nm, so maximum power (non-fatigued) at 40 rpm would be considerably less than 590 W. It seems highly unlikely that the subjects of Marcora and Staiano would have been able to Fig. 1 The linear torque-cadence relationship is plotted with solid symbols and the straight lines represent the regression lines, with r 2 = 0.9777 in the non-fatigued state and r 2 = 0.9784 in the fatigued state. Open symbols represent calculated power (on right axis) corresponding to the same measures as the torque values. Clearly, there is a parabolic power-cadence relationship Communicated by Susan Ward.