Video-oculography (VOG) is a non-invasive method used for the eye tracking detecting. A most important work of VOG is to accurately estimate the pupil center. However, VOG cannot acquire eye movement when the eye blinks. A Grey forecasting model was proposed to predict the pupil center for the blinking frame of VOG video. The Grey forecasting model used GM(1,1) model to carry on the modeling. Then an experiment is provided. The experiment results show that the predicted data of the Grey forecasting meet the characteristics of the patients with nystagmus. The Grey forecasting model is a viable means of accurately predict the pupil center for blinking in VOG.