Having noted the production of propulsive torque throughout much of the upstroke, we want to reemphasize the relative unimportance of the upstroke for doing work when compared to the downstroke. The importance of the upstroke at the high workload was that, by decreasing the negative torque, the cyclists effectively decreased the work that must be done during the downstroke of the other leg, thereby doing less positive work for the same power output. The positive work done during the downstroke was 98.6% at the low workload and 96.3% at the high workload. Although elite subjects were able to reduce the work requirements of one leg during the downstroke by eliminating negative torque production or by producing some slight positive torque during the upstroke with the other leg, they did not use a strategy of producing significant positive work during the upstroke at the higher workload. Thus there is little evidence that feedback devices used to increase pulling up during the upstroke (e.g., Briggs et al., 1989) would help improve the performance of elite cyclists. Nevertheless, since the amount of negative torque produced during the upstroke by the elite subjects in our study appeared significantly less than the amount produced by less skilled subjects in other studies (Lafortune and Cavanagh, 1983; Ericson & Nissel, 1988), it is possible that less skilled cyclists might benefit from learning to reduce negative torque during the upstroke, without attempting to pull up significantly.