The invention belongs to the technical field of eye movement tracking, and discloses an eye movement interaction method, system and device based on eye movement tracking technology, and the method comprises the steps: setting a passive adsorption gaze cursor or an eye movement interaction intention of a sensing area and predicting an active adsorption gaze cursor to select a target; setting corresponding sensing areas, namely effective clicking areas, for different targets, when the cursor makes contact with or covers the sensing area of a certain target, whether eye movement behaviors such aseye tremor exists or not, detecting whether the glancing distance exceeds a threshold value or not at the same time, and absorbing or highlighting then the target object. A machine learning algorithmis adopted to train the eye movement behavior data of the user, the data is filtered, processed and analyzed, an eye movement behavior rule is trained, and a subjective consciousness eye movement interaction intention model of the user is obtained. Through the method, the stability and accuracy in the eye movement interaction process are improved, and the user experience of eye movement interaction is improved.