Forcemyography (FMG) is a useful method to record real-time body motions, which has application potentials for human-machine interactive control. The FMG registers the change of force distribution in the normal direction on muscle surface during limb movements, and the body motions can be recognized by decoding the FMG patterns. In this study, we used FMG to record upper-limb movements and evaluated the influence of different configurations of signal channel and feature on motion classification performances. A four-channel wearable FMG acquisition system was developed to record seven upper-limb movements on each of six able-bodied subjects. The preliminary results showed that the signal channel number has significant influence on motion classification performance; however, the influence of signal feature number on motion classification was insignificant. In addition, the influence of channel combination and feature combination were also discussed in this paper. This work would support the application potential of FMG for body motion recording and may provide useful instructions for the application of FMG in human-machine interactive control.