Abstract Sizing of building air-conditioning systems is a critical issue in design practice concerning the building energy consumption in operation and risk of being undersized. In current practice, chillers and pumps are often oversized due to the rough consideration of uncertainties using safety factor to avoid the risk of being undersized, which results in significant energy waste in operation. In addition, the current design and commissioning do not provide means and flexibility for the air-conditioning systems to minimize their energy consumption when the systems are found oversized in operation. This paper presents a novel design and commissioning approach, consisting of probabilistic optimal design and on-site adaptive commissioning, for building air-conditioning systems to maximize their energy savings in operation. The probabilistic optimal design of an air-conditioning system involves two parts. One is probabilistic optimal design of chillers considering uncertainties, one is probabilistic optimal design of water circulation system considering uncertainties and the flexibility of on-site adaptive commissioning. Monte Carlo simulation is used to quantify uncertainties in system design and operation process. The on-site adaptive commissioning method has alternative commissioning schemes developed to maximize the energy saving based on the actual situation. A case study is performed to test and validate this new design and commissioning approach. Results show that about 20% energy saving could be achieved when the system is oversized by 20%, compared to conventional design and commissioning methods.