Controlling solar furnace systems presents significant challenges due to their nonlinear dynamics and uncertainties in model parameters. Therefore, this paper provides a comprehensive study of four predictive control strategies specifically tailored for solar furnaces: linear generalized predictive control (GPC), nonlinear GPC (NGPC), nonlinear model predictive control (NMPC), and practical NMPC (PNMPC). The primary objective is to address practical issues in solar furnaces, including nonlinear behavior, measured and unmeasured disturbances, and optimal control actions to enhance control performance and reliability in thermal resistance trials. Using real data from an actual solar furnace facility, the control strategies are evaluated in a simulation environment, considering various aspects such as control performance, computational burden, and robustness. Among the strategies, PNMPC proves to be the most promising, attending a compromise between control performance and computational cost. It exhibits a small error-index and significantly shorter processing time (20 times less) compared to NMPC in the simulated test. Consequently, PNMPC is implemented in the existing solar furnace SF60 in Plataforma Solar de Almería, Spain. Real-world results demonstrate the effectiveness of PNMPC in controlling the sample temperature during thermal stress trials in the solar furnace. The controller successfully handles system constraints and performs exceptionally with no steady-state error. As a result, the research outcomes provide suitable solutions to meet high-criteria requirements in thermal stress experiments in solar furnace systems. Furthermore, this study’s findings advance the control engineering field in solar furnace systems, facilitating the transition towards sustainable and efficient use of solar energy.