This study aimed to develop an office-zone grey-box model for a university campus. Based on this, model-based predictive control was simulated. An experiment was conducted using eight temperature sensors in an office zone (25 m²) that accommodates four occupants. Data on the indoor temperature and heating system operating schedules were collected at one-minute intervals. Outdoor temperature and solar radiation data provided by the Korea Meteorological Administration were incorporated into the modeling process in the Matlab environment. The results of the grey-box modeling, conducted over 36 days, revealed satisfactory prediction performance, with an RMSE of 0.7℃. Model-based predictive control simulation was then performed using the developed model. Heating by occupants or by automatic operation according to the schedule did not satisfy the comfort conditions at the start of occupancy, while MPC accurately predicted and controlled the indoor temperature within the comfort bounds. Compared to parametric simulations using different pre-heating strategies of feedback control, MPC showed excellent energy cost savings while maintaining comfort during the heating period.