The fatigue characteristics of paralyzed muscles were investigated during dynamic cycling movement induced by functional electrical stimulation (FES). The peak-to-peak (PTP) amplitude of stimulus-evoked electromyogram (EMG), after suppression of stimulus artifact, was adopted as fatigue indicator. Compared to static contraction, the effects of dynamic movement factors on the stimulus-evoked EMG, such as the intermittent stimulation, joint angle, and contraction speed, were first evaluated in separate experiments. The results of isolated tests laid the foundation for interpreting the data obtained in two FES-cycling experiments, performed under maximum stimulation or in controlled cycling speeds. The effects of intermittent stimulation and joint angle caused periodic changes in PTP amplitude which can be alleviated by averaging the PTP amplitude of one cycle. Under the same stimulation intensity, our results indicated that slower muscle contraction speed would have larger PTP amplitude and vice versa. For the limited number of subjects with paraplegia studied, our results showed that the use of EMG PTP as reliable muscle fatigue indicator during dynamic movement is only valid at the same cycling speed or corresponding contraction speed. The decline of the PTP amplitude decreased with the decay of muscle force can be observed during cycling movement; however, reduction of cycling speed had the opposite effect on PTP amplitude. Observations from the hyperbolic modeling of fatigue process demonstrated that the EMG PTP of a fatigued muscle under dynamic movement decreased at a slower rate.