This paper presents an experimental and numerical study of predictive control of a hybrid ventilation system in an institutional building with emphasis on thermal comfort. A 17–story high institutional building with a large amount of exposed structural concrete is used as a case study to test different control strategies. A model of the entrance region of a typical floor is developed, calibrated from full–scale tests, and validated on a different set of measurements. It is used to estimate the heat removed from the concrete floor and the impact on thermal comfort. The system is currently operating based on heuristic control, with an exterior air temperature setpoint above which the air is allowed into the corridors, which are buffer spaces. With input measured data from 12 typical days, the model is used to compare this strategy to reactive and predictive control strategies. A control strategy based on data of the previous hour (reactive) is expected to quintuple the energy savings potential, but reduce thermal comfort. A predictive control strategy satisfies thermal comfort, but also increases energy savings potential. Lowering thermal comfort criteria during the night in order to precool the building further increases energy savings potential. A prediction horizon of 1 h and a control horizon of 15 min provide satisfactory thermal comfort, while keeping the computational time and forecast uncertainty to reasonable levels for a typical floor. This result provides the basis for the development of an integrated predictive control strategy for the whole building.