Appropriate design provides the cooling system to achieve good performance with low energy consumption and cost. Conventional design method in heating, ventilation and air-conditioning (HVAC) field usually selects the cooling system based on certain cooling load and experiences. The performance of the selected cooling system may deviate from the expectations due to cooling load uncertainty. This paper proposes a novel design method to obtain the robust optimal cooling systems for buildings by quantifying the uncertainty in cooling load calculation and equipment reliability in operation comprehensively. By quantifying the cooling load uncertainty with Monte Carlo method and chiller reliability using Markov method, the robust optimal cooling system is obtained with minimum annual total cost. By applying the new method in the design of the cooling system for a building, its function and performance as well as potential benefits are demonstrated and evaluated. Results show that the proposed method can obtain the optimal cooling systems with low cost and high robustness and provides a promising means for designers to make their best design decisions based on quantitative assessment according to their priority. (C) 2015 Elsevier Ltd. All rights reserved.