This study aims to verify maneuverability of a man-machine interface (MMI) based on electromyogram (EMG) by comparing the MMI with an ordinal haptic-type MMI. It is required for a MMI to let a target system work as the operator intends, and no time delay between the operator intention and the machine response is ideal. However, machines have dynamics and time delay such as phase-lag is inevitable. This study, hence, proposes a EMG-based MMI utilizing Zero-Phase Error Tracking Controller (ZPETC) to reduce the time delay. Consisting ZPETC, a nonlinear ARX model plays an important role to estime the physical exertion from EMG. The effectiveness of the MMI is verified through an experiment, ball-hitting task. As the experimental result, the average success rate concerned with the task using the haptic-based MMI is 46.7 %, and the average success rate using the proposed MMI is highly improved up to 66.7 %. By this experiment, the average success rate of the task by MMI using a haptic device was set to 46.7 [%], and the average success rate of the task by MMI using myoelectric interface was set to 66.7 [%].