In this study, an effective noncontact and nonattached technique that is based on electrostatic induction current generated during walking motion is proposed for the detection and assessment of human physical activity. In addition, a theoretical model is proposed for the electrostatic induction current generated owing to variation in the electric potential of the human body. The proposed electrostatic induction current model is compared with the theoretical model, and the proposed model is shown to effectively explain the behavior of the electrostatic induction current waveform. The normal walking motions of daily living are recorded with a portable sensor located in a regular house. The obtained results show that detailed information of physical activity such as a gait cycle can be estimated using our proposed technique. Additionally, the walking signal was measured when the subject walked with the ankle and knee fastened to a splint with bandages to simulate a limp. Therefore, the proposed technique, which is based on the detection of signal generated during walking, can be successfully employed to assess human physical activity.