Keeping balance is the premise of human walking. ZMP(zero moment point) is a point where total torque achieves balance. It is an important evaluation parameter of balance ability in walking, since it can be used to better reflect the dynamic balance during walking. ZMP can be used in many applications, such as medical rehabilitation, disease diagnosis, treatment and etc. In this paper, wearable inertial sensors system based on MEMS is used to measure ZMP(zero moment point) during walking, which is cheap, convenient, and free from the restriction of lab. Our wearable ZMP measurement system consists of inertial measurement subsystem and PC real-time monitoring station. Inertial measurement subsystem includes 9-axis inertial sensing nodes, the body communication network and the central node. Inertial sensing nodes are mounted on different parts of the body to collect body posture information in real-time, and then the best estimation of current posture are obtained by Kalman filter. The data from sensors is aggregated to the central node through the CAN bus, and then ZMP is calculated. Finally, it can be showed in the PC monitoring station. Experiments prove the system can achieve real-time ZMP detection during walking. Keeping balance is the premise of human walking. ZMP (zero moment point) is a point where total torque achieves balance. It is an important evaluation parameter of balance ability in walking, since it can be used to better reflect the dynamic balance during walking. ZMP can be used in many applications, such as medical rehabilitation, disease diagnosis, treatment and etc. In this paper, wearable inertial sensors system based on MEMS is used to measure ZMP (zero moment point) during walking, which is cheap, convenient, and free from the restriction of lab. Our wearable ZMP measurement system consists of of inertial measurement subsystem and PC real-time monitoring station. Inertial sensing nodes are 9-axis inertial sensing nodes, the body communication network and the central node. mounted on different parts of the body to collect body posture information in real-time, and then the best estimation of current posture are obtained by Kalman filter. The d ata from sensors is aggregated to the central node through the CAN bus, and then ZMP is calculated. Finally, it can be showed showed in the PC monitoring station. Experiments prove the system can achieve real-time ZMP detection during walking.