Improving the intuitiveness of the interaction between human and machine is an important issue for powered lower-limb prosthesis control. In this research, we aimed to evaluate the potential of using surface electromyography (EMG) signals measured from transtibial amputees' residual muscles to directly control the position of prosthetic ankle. In this research, one transtibial amputee subject and five able-bodied subjects were recruited. They were asked to control a virtual ankle to reach different target positions. The amputee subject finished these tasks in an average time of 1.29 seconds for different target positions with the residual limb, which was comparable with that using the amputee's sound limb and those with able-bodied subjects' dominant legs. Due to human's strong adaptability, the amputee subject was able to adapt to the control model trained one day before or trained in a posture which was different from that during performing control tasks. These results validate the promise of using surface EMG signals to volitionally control powered transtibial prostheses.