The object of this study was to compute the mechanical power of the resultant braking force during an actual propulsion cycle with a manual wheelchair on the field. The resultant braking force was calculated from a mechanical model taking into account the rolling resistances of the front and rear wheels. Both the resultant braking force and the wheelchair velocity were not constant during the propulsion cycle and varied according to the subject's fore-and-aft and vertical movements in the wheelchair. These variations had logical repercussions on the braking force mechanical power, which ranged from 20.6 to 34.5 W (mean = 29.6 W) during the propulsion cycle. The mechanical power was also calculated from the conditions of a classical drag-test, by the product of the cycle mean velocity and a constant braking force corresponding to a 60 % rear wheels distribution of the subject-and- wheelchair's weight. This second mechanical power (32.4 W) was 10 % higher than the average of the instantaneous power. Beyond the need of a clear definition of the two phases of the propulsion cycle, this study showed that the assumption on wheelchair locomotion usually admitted on laboratory ergometers cannot be applied in field studies, and that the kinetic energy variations during the cycle propulsive phase should be considered for evaluating the subject's mechanical work and power.