Energy saving potential of radiant floor cooling combined with underfloor ventilation (RFCUV) system has been restricted by its control problems. Existing conventional control methods of radiant cooling system have several disadvantages, such as control lag, poor control performance, and low economy, etc. The objectives of this study were to: (i) build a dynamic simplified model and validate its precision experimentally; (ii) implement advanced model predictive control (MPC) on RFCUV system; and (iii) demonstrate MPC control performance by comparing with existing conventional proportional-integral-derivative (PID) experimentally. Experimental results indicated that under experimental step setpoint variations, the adjusting time of indoor air temperature or operative temperature was only 12 min with MPC controller, and was 30 min with PID controller; it took only 1 min to reach recommended thermal comfort range with MPC controller, and 17 min with PID controller. During 9:00 to 17:00 in typical design day of Nanjing city, compared with PID controller, MPC controller yielded 17.5% energy saving when maintaining equal or better indoor comfort. Thus, compared with PID, MPC demonstrated the advantages of rapid responses, good stability and excellent energy saving effect in RFCUV system. (C) 2019 Elsevier Ltd. All rights reserved.