The invention discloses an epileptic seizure monitoring method and system based on depth data. The method comprises steps of: acquiring a data source of an epileptic by using a depth sensor; determining whether the epileptic is likely to have as epileptic seizure by analysis of a resultant accelerated speed threshold value of joints; tracking the loci of the joints by using an optical flow method, computing sampling point speed to achieve feature extraction, acquiring a feature capable of describing the drastic vibration of the arms and the legs of the epileptic when the epileptic has an epileptic seizure; identifying whether the current epileptic has an epileptic seizure by using a support vector machine classification method, and if yes, enabling a loudspeaker to give alarm sound, enabling an LED to twinkle, and automatically sending a short message or email via GSM and 4G mobile communication to notify family members. The system comprises a depth data acquisition module, a preanalysis module, an identification module, a data storage module, and an alarm module. The method and the system may accurately and timely monitor the vibration of the arms and the legs of the epileptic having an epileptic seizure, and are insusceptible to environmental changes such as light or the like, fast in detection speed, and good in robustness.